Micromorphological characterization of polymer-oxide nanocomposite thin films by atomic force microscopy and fractal geometry analysis

Progress in Organic Coatings 89 (2015) 50–56

Contents lists available at ScienceDirect

Progress in Organic Coatings journal homepage: www.elsevier.com/locate/porgcoat

Micromorphological characterization of polymer-oxide nanocomposite thin films by atomic force microscopy and fractal geometry analysis S¸tefan T¸a˘ lu a , Niranjan Patra b,1 , Marco Salerno b,∗ a Technical University of Cluj-Napoca, Faculty of Mechanical Engineering, Department of AET, Discipline of Descriptive Geometry and Engineering Graphics, 103-105 B-dul Muncii St., Cluj-Napoca 400641, Cluj, Romania b Istituto Italiano di Tecnologia, Nanophysics Department, Scanning Probe Microscopy Lab, via Morego 30, I-16163 Genova, Italy

a r t i c l e

i n f o

Article history: Received 16 February 2015 Received in revised form 28 May 2015 Accepted 27 July 2015 Keywords: Atomic force microscopy Polymer-oxide nanocomposites Thin films Fractal analysis Surface roughness Micro-morphology

a b s t r a c t The purpose of this study was to evaluate the 3-D surface micromorphology of polymer-oxide thin films spin-coated from a composite of poly-methyl-methacrylate as the matrix and elongated titania nanorods as the filler particles. The surfaces of these composite films were investigated by atomic force microscopy and characterized by fractal geometry analysis. The effect of increasing loading of the fillers between 0 and 30% by weigth relative to the matrix was assessed. An increasing roughness was observed, with typical emergence of protruding ripples progressively extending into larger stripes. The amplitude parameters of the surfaces were determined by analysis of the height distributions. The fractal analysis of roughness revealed that the films have fractal geometry. Triangulation method based on the linear interpolation type was applied to determine the fractal dimension. A connection was observed between the surface morphology and the physical properties of the coatings as assessed in previous works. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The progress in control of polymeric materials and their hybrid composites with inorganic particles during the last twenty years has led to their use in a number of applications [1–3]. Composites of organic matrix and nanoparticles as the fillers are used in many devices, from OLEDs to photovoltaic solar cells and gas sensors [4], as well as in structural materials [5] for aeronautics, automotive, biomedical [6] and dental restoration [7]. The nanocomposites are used in bulk or as coatings. In both cases, but particularly for the thin coatings, homogeneous distribution of the filler particles inside the matrix is required, even at high loading, without appearance of large aggregates, which are often accompanied by phase segregation [1–3,5]. Phase segregation is particularly detrimental not only for the expected bulk properties of the composite, which are the result of careful design of the matrix-filler mixture [8,9], but also for their surface. In fact, may the composite be a coating or

∗ Corresponding author at: Istituto Italiano di Tecnologia, Nanophysics Department, Scanning Probe Microscopy Lab, via Morego 30, I-16163 Genova, Italy. Tel.: +39 010 71781444; fax: +39 010 720321. E-mail address: [email protected] (M. Salerno). 1 Currently at Imperial College, Faculty of Engineering, Department of Materials, South Kensington Campus, London SW7 2AZ, UK. http://dx.doi.org/10.1016/j.porgcoat.2015.07.024 0300-9440/© 2015 Elsevier B.V. All rights reserved.

a bulky piece, the surface roughness is one of the most important properties to be controlled, affecting e.g. the lifetime of lubricating surfaces or the wetting properties of self-cleaning materials [10]. In most cases a high roughness is undesirable (e.g. for optical components or surfaces exposed to environmental bacteria that should not adhere [11,12]), but in some cases controlled roughness can be of help (e.g. for orthopedical and dental implants, where new tissue on-growth is fostered by tuned height and spacing of surface features [13]). A common polymer used often as the model matrix in experimental nanocomposites is poly-methyl-methacrylate (PMMA) [3,4,6]. An interesting candidate as the filler nanoparticle material is the semiconducting oxide of titanium, TiO2 , namely titania. This is of interest for applications in photonics and photoelectronics, thanks to its light-guiding high refractive index (n = 2–2.3 in the visible) [14]. Additionally, the photo-chemical properties of titania make it useful for catalytic degradation of pollutants [15] and for controlled wetting on UV irradiation [16], which suggest using its composites for lithographic fabrication of microfluidic biomedical devices. We studied nanocomposite thin films of titania nanorods (NRs) dispersed in PMMA by means of atomic force microscopy (AFM). Fractal geometry is an advanced mathematical tool that has been used for quantifying the irregular complex structures and patterns across many spatial or temporal scales, relevant to

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56

applications in surface engineering research [17–20]. Whereas to our best knowledge no multifractal geometry study of the 3-D surface micromorphology of polymer-oxide nanocomposite coatings exists in the available literature, we think that this analysis can give deeper insight in the complexity of the topographical patterns emerging in similar materials. Therefore, the purpose of this study was to investigate the 3-D surface micromorphology of our polymer-oxide nanocomposite thin films.

51

Statistical analysis between Ds values from different images (N = 5) was performed using the SPSS 14 for Windows (Chicago, Illinois, USA). One-way ANOVA was used to test the differences between the two groups with Scheffé post-hoc tests for multiple comparisons, considering statistically significant the differences with a P value <0.05.

3. Results and discussion 2. Materials and methods 2.1. Materials The fillers were prolate titania NRs, of crystalline character (anatase in phase), synthesized in toluene as described elsewhere [21]. The typical dimensions were 30 nm in length and 4 nm in diameter. The NRs presented a coating of oleic acid, working as a surfactant, to prevent aggregation. The suspension of NRs was mixed with PMMA (Aldrich, Italy) of Mw  = 120 kDa, at different relative concentration by weight , from 0 (bare polymer) up to 30%. The absolute concentration of PMMA was changed such as to maintain approximately the same thickness of the resulting film, which was always ∼100 nm. The mixture was spin coated on glass at fixed conditions of 2000 RPM for 1 min and let to dry in ambient air. 2.2. Methods 2.2.1. AFM measurements All AFM images of the samples were acquired in ambient conditions (21 ± 1 ◦ C) and (44 ± 2% RH) with an MFP-3D Bio (Asylum Research, USA) working in contact mode. The probes used were CSG10 (NT-MDT, Russia), with nominal spring constant and tip apex diameter of ∼0.1 N/m and ∼10 nm, respectively. The scans were performed on areas of 10 × 10 ␮m2 , collecting images with 512 × 512 pixels. The measurements were repeated for four times for each sample on different reference areas to validate the reproducibility of the typical features. All images have been modified only with plane fitting for background subtraction, and with no filter or other non-linear treatment (such as e.g. line-by-line flattening). 2.2.2. Analysis of the AFM images For parametric description of the 3-D surface, the images have been analyzed with dedicated AFM software Gwyddion [22]. Detailed information on the quantitative parameters extracted from the AFM images of topography can be obtained by the respective definitions, according to published standards [23]. The fractal analysis was applied to the AFM images also by means of the above-mentioned freeware program Gwyddion. The triangulation method was used, with a linear interpolation type. In this method the fractal dimension Ds is obtained directly from boxcounting of the surface with triangles. At each step i a grid of linear size units li is placed on the surface, which defines the position of the vertices of triangles, tiling the surface texture and presenting different angles of inclination with respect to the xy base-plane. The areas of all triangles are calculated and summed to obtain an approximation of the surface area S(li ). The initial grid size is l1 = L, same as the whole image scan size (in our case L = 10 ␮m). li is progressively decreased by a factor 2, so that li = l1 /2i−1 , and the process continues until li corresponds to the single pixel size p (in our case p = L/512 ≈ 20 nm). Thus, the number of li values is n such that p = L/2n−1 , which means n = 10. In the used program the S(li ) is then plotted versus ln(p/li ) and the slope of this curve identifies Ds − 2.

A set of 3-D topographic AFM images selected to be representative of the nanocomposite films with different NR filler loading  is shown in Fig. 1, with  = 0, 5, 10, 20 and 30 wt%, for panels (a)–(e), respectively. The surfaces are rendered in 3-D perspective view, with the vertical (height) scale displayed in color coding, according to the palette presented on the respective right hand side, and the scale chosen to stress the surface texture on each image. Clearly, the bare PMMA film ( = 0) looks quite flat and smooth, while in the nanocomposite films features appear in the form of round grains ( = 5%), progressively coalescing into connected ridges of increasing width. In Fig. 2 the peculiar type of plots used for the analysis carried out in this work are shown, for the single case of the representative AFM images presented in Fig. 1. In the different rows, the plots corresponding to the images Fig. 1a–d with increasing NRs filler loading  = 0–30% are shown. In the two columns, on the left the distributions of heights appear, whereas on the right the plots used for calculating the respective fractal dimensions Ds appear. These latter plots report the data points, and the respective linear fits, of the approximated surface area S(li ) resulting from the triangulation method, versus a parameter describing the finesse of the grid used for triangulation, in pixel size units and logarithmic scale, namely ln(p/li ). From the height distributions as in Fig. 2a, c, e, g and i, the values of height range (i.e. maximum–minimum), median, average roughness Sa , root mean square roughness Sq , skewness Ssk and kurtosis Sku have been calculated. After repeating this calculation for all the different sets of AFM images (N = 5), the resulting values have been listed in Table 1, in the form of mean ± one standard deviation. In Fig. 3 these results are presented in the form of plots versus the NRs loading . It can be seen from Fig. 3a that all the height parameters (height range, median, Sa and Sq ) increase monotonically with , and in particular the roughness parameters Sa and Sq increase along an almost straight line with a weak tendency to saturation and only slightly higher slope for Sq (∼1.8 vs ∼1.5 nm/%). The linear increase in roughness with ␾ observed here is consistent with previous reports on similar samples with NRs of both anatase [24] and brookite [25]. In this work, in addition to the width Sq we also observed the higher moments of height distribution of our samples. The skewness qualifies the symmetry, as negative values indicate predominant valleys and positive ones indicate predominant peaks. From Fig. 3b it appears that, while the starting surface (bare PMMA) is balanced in peaks and valleys, the type of roughness emerging on NRs loading is more of peak–type, with strongest peakedness for the 5% sample. Kurtosis instead qualifies the type of protruding features providing the roughness, since for spiky (bumpy) surfaces it is >(<) 3. Again the 5% surface is clearly different from the others, and it qualifies as spiky. This is the sample where roughness starts to arise as the result of the presence of single NRs or small aggregates. On the contrary, all the other surfaces present rather bumpy-type peaks. As compared to the standard quantities reported in Table 1 and Fig. 3, fractal geometry analysis may offer new parameters for characterizing the morphology of complex objects [17–20]. In particular, the surface area S of a surface fractal object increases with its size (equivalent radius) r according to S ∝ rDs , with Ds surface

52

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56

Fig. 1. Representative 3-D topographic AFM images of the polymer-oxide nanocomposite thin films surfaces, with different loading  of titania NR fillers, in wt% relative to the PMMA matrix: (a)  = 0%, (b)  = 5%, (c)  = 10%, (d)  = 20%, and (e)  = 30%. Scan area is 10 ␮m x 10 ␮m.

fractal dimension. Ds is a non–integer value within the range 2 ≤ Ds ≤ 3 and is a generalization of the intuitive notion of topological dimension which is used to describe the surface complexity [26].

The fractal analysis of all the polymer–oxide nanocomposite thin films surfaces as resulting from AFM images as in Fig. 1 has been carried out by the triangulation method [27] based on the linear interpolation type. The results of fractal analysis, including the

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56

53

Fig. 2. Height distributions on the left (a, c, e, g, i) and plots used to calculate the fractal dimension Ds on the right (b, d, f, h, j) for the representative samples of Fig. 1, for different titania NR filler loading  of 0 (a, b), 5 (c, d), 10 (e, f), 20 (g, h) and 30 (i, j) wt%, on the different lines.

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56

54

Table 1 Values of the parameters describing the height distributions of the all the AFM images (N = 5) as those in Fig. 1. Sample  (wt%)

Parameter

Height range (nm) Median height (nm) Sa (nm) Sq (nm) Ssk ( ) Sku ( )

0

5

10

20

30

4.6 ± 3.0 2.1 ± 2.0 0.25 ± 0.18 0.31 ± 0.30 0.04 ± 0.03 0.35 ± 0.10

139 ± 32 19.5 ± 8.6 11.7 ± 7.3 16.2 ± 9.3 1.93 ± 0.75 4.49 ± 2.15

160 ± 41 72.6 ± 32.0 21.6 ± 10.5 25.9 ± 11.0 0.30 ± 0.18 −0.66 ± 0.24

212 ± 80 63.6 ± 58.9 35.1 ± 21.8 42.4 ± 18.4 0.70 ± 0.22 −0.42 ± 0.13

363 ± 78 164 ± 39 44.7 ± 18.5 57.3 ± 22.1 0.30 ± 0.13 0.34 ± 0.16

Fig. 3. Plots of the statistical quantities extracted from the distribution of heights of the AFM images of all sets of samples (N = 5, for example see left column in Fig. 2) versus the titania NR filler loading . (a) quantities describing the distribution size, namely the height range, height median, average roughness Sa and root mean square Sq ; (b) quantities describing the distribution shape, namely the skewness Ssk and kurtosis Sku .

Table 2 Fractal dimensions Ds with mean coefficients of correlation R2 , for all the AFM images obtained for the polymer-oxide nanocomposite thin films with different  (N = 5), similar to those ones presented in Fig. 1. # means that for the special case of bare polymer a 3 × 3 spatial low pass filtering has been applied on the images before calculation. *: statistically significant difference with all the other samples. Parameter

Ds R2

Sample  (wt%) 0

5

10

20

30

2.43 ± 0.12# 0.992

2.37 ± 0.15 0.993

2.35 ± 0.11 0.992

2.39 ± 0.10 0.994

2.16 ± 0.12* 0.993

coefficients of correlation (R2 ), are presented in Table 2. Actually, for fractality of a surface to be defined, the commonly accepted rule of thumb is that the analysis should extend for at least three decades. In fact, for the data points fitted in the plots of the right columns of Fig. 2, the nine halving steps of li make the number of explored decades equal to log10(29 ) ≈ 2.7, which hardly approximate the above value. We are aware of this limitation, which is intrinsic in the selected AFM technique, whose images carry a limited amount of spatial information, due to the slow acquisition. More accurate analysis would require 1024 × 1024 pixels, which will double further the acquisition times, or different techniques such as SEM, but with the loss of direct 3-D information. The means (N = 5) of R2 of linear fits at different  for all the sets of AFM images were all greater than 0.991, representing a good linear correlation. The Ds values are reported as mean ± one standard deviation. The higher fractal dimension Ds of 3-D surfaces appears for the bare polymer film, whereas the lower one appears for the films with highest loading  = 30%. Since Ds is a measure

of the global scaling property, we can say that, while almost flat, the bare polymer surface is the one with the apparently highest self–affinity [26]. In fact, ideally the bare polymer should be perfectly flat and smooth, which would give a value of Ds = 2. However, while being the smoothest of all samples, a base roughness also exists for the polymer film, which can be ascribed to very low fluctuations in thickness on the nanometer scale (see Sq for the height spread in either Fig. 2a or Table 1). Actually, even a random noise surface would present a comparatively high Ds value, higher than the real samples with ␾ = 5–30%. The point is that Ds alone does not fully describe the roughness, but is rather a measure of roughness texture, while at least one amplitude parameter such as Sq is required, for a more realistic description. Therefore, the Ds calculated for the raw data of sample  = 0 is artificially overestimated. It is sufficient to apply a Gaussian 3 × 3 kernel filter in the real space to the images with ␾ = 0, to depress the respective Ds to 2.43 ± 0.12, which makes it no more statistically different from all the other samples but  = 30. When the same filtering was applied to the images of the other samples, overall a much lower decrease in Ds was observed (–1% as compared to the –4% for sample  = 0). For the samples representing the composite films ( > 0), the Ds values for  = 5, 10 and 20 are roughly the same, and indeed there appears no statistically significant difference among them. On the contrary, sample  = 30% is statistically different with respect to all of them. We can say that the fractal dimension Ds stays roughly the same for increasing , up to a limit where is decreases, between  = 20 and 30%, as the likely aggregation of NRs under the surface gives rise to simpler texture (large stripes), despite the higher roughness amplitude Sq .

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56

In our former studies involving similar films of titania NRs we have demonstrated that an appropriate quantification of the surface roughness is important, as this is associated with both the contents and dispersion of NRs, which in turn affect all the properties of interest, namely the stiffness [24,25], the viscoelasticity [28], the thermal stability [29] and the wettability [16]. For example, for anatase NRs the elastic modulus was observed to increase linearly with  up to 30%, while the hardness increased stepwise at around 15%, and this threshold was the same at which the glass transition temperature Tg increased up to a saturation level [24]. For the brookite NRs, which had aspect ratio of ∼20 instead of ∼7, and thus 30-fold higher Sq , modulus and hardness showed a maximum for  ≈ 10% and decreased again at higher loading due to aggregation, which was associated with higher surface roughness Sq . On the other hand, significant aggregation was found also for short NRs in films drop cast on substrates, as observed by cross section TEM views of similar films [28], at a lower threshold of  = 2%. The improvement in thermal stability of these nanocomposites (Tg increase of 10–15 ◦ C vs PMMA) was assigned to the NRs preventing fast heat diffusion and limit the matrix degradation [29], called a barrier effect, somewhat similar to that observed for the mechanical strength of composites (crack propagation stop). The morphological study of these types of samples was previously limited to the observation that with increasing loading  a roughly linear increase in surface roughness occurred, solely described with RMS parameter Sq . For long NRs a local maximum in modulus and hardness was observed at loading as low as 5–10%, while for short NRs a stable level in mechanical properties was reached at ∼15% loading. This effect was ascribed to the long NRs providing a stronger improvement at low loading, yet giving rise more easily to aggregates and a loss of most previous improvement at higher loading. The system is complicated by the presence of ∼26 wt% OLAC content as a surfactant, coating the NRs, and 4 wt% water, acting as a plasticizer, as demonstrated by TGA measurements of both bare NRs and nanocomposites [24]. In a work made with short NRs reaching higher loading up to ∼70% [16], the occurrence of aggregates was also observed by assessing the optical properties, which resulted in increased light scattering at high loading, with a critical concentration for clustering ∼20%. The roughness in that case also affected the wetting properties, with water contact angle (CA) increasing from 70◦ at higher rate up to the 20%, and the staying constant at 100◦ . From the independent measurements of CA it was also possible to assess the coverage of the wetting water, according to Cassie–Baxter model [30], thus localizing the likely air–pocket positions. In our analysis we show that deeper insights can be given in the characterization of the surfaces of these nanocomposite coatings as obtained by AFM 3-D imaging, and possibly a correlation with the physical effects of loading can be established, upon careful evaluation. In particular, extended analysis of AFM images can be achieved by both extending the list of roughness amplitude parameters, as obtained by most common computer programs, and by assessing the fractal dimension Ds of the surface features. Whereas Ds rather describes the texture only and should be taken with care (see the above discussion of highest apparent Ds obtained for the ‘flat’ polymer), a combination of the two types of information (amplitude parameters in z, and Ds for the description of texture pattern in the xy plane) can represent a powerful tool for better evaluation of the effects of sample micro- and nano-scale morphology.

4. Conclusions This study confirms the previous results obtained on thin films of nanocomposites of PMMA and titania NRs, which demonstrated that the sample surface morphology depends on the material

55

composition, namely the NRs loading level . This, in turn, is known to affect several physico–chemical properties, such as stiffness, viscoelasticity, thermal stability and wetting. Our results suggest that advanced statistical surface roughness parameters such as higher moments of the height distributions and fractal dimension may provide additional insight into the nature of the polymer–oxide nanocomposite thin film, and better quantify and identify the different 3-D geometrical patterns arising during casting. These results can be used to develop mathematical theoretical models which help to study structure, simulation of dynamics at interfaces and thermodynamics processes at nanometer level. Financial disclosure Neither author has a financial or proprietary interest in any material or method mentioned. Competing interest The authors declare that they have no competing interests. References [1] F. Hussain, Review article: polymer-matrix nanocomposites, processing, manufacturing, and application: an overview, J. Compos. Mater. 40 (2006) 1511–1575, http://dx.doi.org/10.1177/0021998306067321 [2] R. Bogue, Nanocomposites: a review of technology and applications, Assem. Autom. 31 (2011) 106–112, http://dx.doi.org/10.1108/01445151111117683 [3] I.-Y. Jeon, J.-B. Baek, Nanocomposites derived from polymers and inorganic nanoparticles, Materials (Basel) 3 (2010) 3654–3674, http://dx.doi.org/10. 3390/ma3063654 [4] D.Y. Godovsky, Device Applications of Polymer-Nanocomposites, 2000. [5] S.C. Tjong, Structural and mechanical properties of polymer nanocomposites, Mater. Sci. Eng. R. Rep. 53 (2006) 73–197, http://dx.doi.org/10.1016/j.mser. 2006.06.001 [6] M.-S. Scholz, J.P. Blanchfield, L.D. Bloom, B.H. Coburn, M. Elkington, J.D. Fuller, et al., The use of composite materials in modern orthopaedic medicine and prosthetic devices: a review, Compos. Sci. Technol. 71 (2011) 1791–1803, http://dx.doi.org/10.1016/j.compscitech.2011.08.017 [7] S.B. Thorat, A. Diaspro, M. Salerno, In vitro investigation of coupling-agent-free dental restorative composite based on nano-porous alumina fillers, J. Dent. 42 (2014) 279–286, http://dx.doi.org/10.1016/j.jdent. 2013.12.001 [8] G.M. Odegard, T.C. Clancy, T.S. Gates, Modeling of the mechanical properties of nanoparticle/polymer composites, Polymer (Guildf) 46 (2005) 553–562, http://dx.doi.org/10.1016/j.polymer.2004.11.022 [9] S.-M. Choi, H. Awaji, Nanocomposites—a new material design concept, Sci. Technol. Adv. Mater. 6 (2005) 2–10, http://dx.doi.org/10.1016/j.stam.2004.07. 001 [10] G. Caputo, B. Cortese, C. Nobile, M. Salerno, R. Cingolani, G. Gigli, et al., Reversibly light-switchable wettability of hybrid organic/inorganic surfaces with dual micro-/nanoscale roughness, Adv. Funct. Mater. 19 (2009) 1149–1157, http://dx.doi.org/10.1002/adfm.200800909 [11] J. Nikkola, Polymer Hybrid Thin-Film Composites with Tailored Permeability and Anti-Fouling Performance, Science - VTT, 2014. [12] S. Cresti, A. Itri, A. Rebaudi, A. Diaspro, M. Salerno, Microstructure of titanium-cement-lithium disilicate interface in CAD-CAM dental implant crowns: a three-dimensional profilometric analysis, Clin. Implant Dent. Relat. Res. (2013) 4–6, http://dx.doi.org/10.1111/cid.12133 [13] M. Salerno, F. Caneva-Soumetz, L. Pastorino, N. Patra, A. Diaspro, C. Ruggiero, Adhesion and proliferation of osteoblast-like cells on anodic porous alumina substrates with different morphology, IEEE Trans. Nanobiosci. 12 (2013) 106–111, http://dx.doi.org/10.1109/TNB. 2013.2257835 [14] X. Chen, S.S. Mao, Titanium dioxide nanomaterials: synthesis, properties, modifications, and applications, Chem. Rev. 107 (2007) 2891–2959, http://dx. doi.org/10.1021/cr0500535 [15] L. Frazer, Titanium dioxide: environmental white knight, Environ. Health Perspect. 109 (2001) A174–A177. [16] F. Pignatelli, R. Carzino, M. Salerno, M. Scotto, C. Canale, M. Distaso, et al., Directional enhancement of refractive index and tunable wettability of polymeric coatings due to preferential dispersion of colloidal TiO2 nanorods towards their surface, Thin Solid Films 518 (2010) 4425–4431, http://dx.doi. org/10.1016/j.tsf.2010.01.041 [17] S¸. T¸a˘ lu, S. Stach, M. Ikram, D. Pathak, T. Wagner, J.-M. Nunzi, Surface roughness characterization of ZnO:TiO2 – organic blended solar cells layers by atomic force microscopy and fractal analysis, Int. J. Nanosci. 13 (2014) 1450020, http://dx.doi.org/10.1142/S0219581X14500203 [18] S¸. T¸a˘ lu, S. Stach, J. Zaharieva, M. Milanova, D. Todorovsky, S. Giovanzana, Surface roughness characterization of poly(methylmethacrylate) films with

56

[19]

[20]

[21]

[22] [23]

S¸. T¸a˘ lu et al. / Progress in Organic Coatings 89 (2015) 50–56 immobilized Eu(III) ␤-diketonates by fractal analysis, Int. J. Polym. Anal. Charact. 19 (2014) 404–421, http://dx.doi.org/10.1080/1023666X.2014. 904149 S¸. T¸a˘ lu, A.J. Ghazai, S. Stach, A. Hassan, Z. Hassan, M. T¸a˘ lu, Characterization of surface roughness of Pt Schottky contacts on quaternary n-Al0.08In0.08Ga0.84N thin film assessed by atomic force microscopy and fractal analysis, J. Mater. Sci. Mater. Electron. 25 (2013) 466–477, http://dx. doi.org/10.1007/s10854-013-1611-6 ˇ D. Dallaeva, S¸. T¸a˘ lu, S. Stach, P. Skarvada, P. Tománek, L. Grmela, AFM imaging and fractal analysis of surface roughness of AlN epilayers on sapphire substrates, Appl. Surf. Sci. 312 (2014) 81–86, http://dx.doi.org/10.1016/j. apsusc.2014.05.086 P.D. Cozzoli, A. Kornowski, H. Weller, Low-temperature synthesis of soluble and processable organic-capped anatase TiO2 nanorods, J. Am. Chem. Soc. (2003) 14539–14548. P. Klapetek, D. Neˇcas, C. Anderson, Gwyddion, (n.d.). Available from: http:// gwyddion.net ISO, ISO 25178-2:2012, Geometrical Product Specifications (GPS) – Surface Texture: Areal – Part 2: Terms, Definitions and Surface Texture Parameters, 2012, http://www.iso.org

[24] N. Patra, A.C. Barone, M. Salerno, G. Caputo, D. Cozzoli, A. Athanassiou, Thermal and mechanical characterization of PMMA-TiO2 , Nanocomposites 67 (2009) 209–215. [25] N. Patra, A. Barone, M. Salerno, G. Caputo, A. Athanassiou, Nanocomposites of Poly(methyl methacrylate) and Brookite Titania Nanorods, 2011. [26] B.B. Mandelbrot, The Fractal Geometry of Nature, Freeman, San Francisco, 1982. [27] C. Douketis, Z. Wang, T.L. Haslett, M. Moskovits, Fractal character of cold-deposited silver films determined by low-temperature scanning tunneling microscopy, Phys. Rev. B 51 (1995) 11022–11032. [28] N. Patra, M. Salerno, P.D. Cozzoli, A.C. Barone, L. Ceseracciu, F. Pignatelli, et al., Thermal and mechanical characterization of poly(methyl methacrylate) nanocomposites filled with TiO2 nanorods, Compos. Part B: Eng. 43 (2012) 3114–3119, http://dx.doi.org/10.1016/j.compositesb.2012.04.028 [29] N. Patra, M. Salerno, M. Malerba, P.D. Cozzoli, A. Athanassiou, Improvement of thermal stability of poly(methyl methacrylate) by incorporation of colloidal TiO2 nanorods, Polym. Degrad. Stab. 96 (2011) 1377–1381, http://dx.doi.org/ 10.1016/j.polymdegradstab.2011.03.020 [30] A.B.D. Cassie, S. Baxter, Wettability of porous surfaces, Trans. Faraday Soc. 40 (1944) 546–551, http://dx.doi.org/10.1039/TF9444000546.