Abstract
Background: Mechanisms of selfassembly/selforganization are fundamental for the emergence of nanoarchitectonic systems composed by elemental units, and it is important to build a theoretical framework for them. Additionally, because the enhanced functionalities of these systems are related to their spatial morphologies, it is necessary to quantify the selforganized design through suited statistical analysis tools.
Results: We have investigated the selfassembly bundling process of nanowires fabricated by metalassisted chemical etching (MACE). First, we have applied theoretical models in order to obtain a quantitative estimation of the driving forces leading to selfassembly. Then, we have studied the surfaces of the nanoarchitectures by means of multifractal analysis. We have found that these systems are not simple monofractals, but that the more complex paradigm of multifractality (different fractal dimensions across different scales) has to be applied to describe their morphology.
Conclusion: The multifractal analysis approach has proven its ability to discriminate among different MACE nanoarchitectures. Additionally, it has demonstrated its capacity to measure the degree of homogeneity of these surfaces. Finally, a correlation between the growth conditions and the capacity dimension of the nanowires was obtained.
Introduction
In the last years, huge progress was made regarding the study and the technological exploitation of materials endowed with new properties deriving from their nanoscale features. In this respect, the field of nanoarchitectonics [1,2] has attracted attention as one of the most promising paradigmatic changes in nanotechnology. In general, the concept of nanoarchitectonics consists in the approach of building up large structures from nanoscaled units by selfassembly. This selfbuilding is driven by the reciprocal interactions among the units, where these interactions are such as van der Waals, electrostatic, magnetic, molecular, and entropic forces [3]. The technological advantage is that in comparison to the nanoscaled units these selforganized assemblies possess new functionalities. Atoms, molecules, or even nanoparticles or nanowires (NWs) can be used as basic units to selfarrange in new wholes.
NWs are among the most widely investigated nanoscaled objects. Especially semiconductor NWs offer the unique promise to boost the performance of semiconductor devices by quantum effects. In this respect, silicon NWs [47] are key elements in the field of nanotechnology, given that they can be integrated in the microelectronic industry, which is mainly Sibased. From a technological point of view, it is essential to explore the possibilities of a largescale fabrication of NWs. The topdown approach [8] represents the main route to achieve this goal, because it allows for waferscale growth by an easy adaptation of microfabrication equipment already available in the industry. The topdown methods involve the use of both dry [9,10] and wet etching [11] to carve nanostructures from a substrate. Metalassisted chemical etching (MACE) [1215] has gained particular attention in this regard, because it is simple, of low cost and versatile. MACE is an anisotropic wet etching technique where the sculpting of the nanostructures is catalyzed by a discontinuous thin film of noble metal deposited on a substrate. The metal works as a local cathode where the reduction of oxidants occurs. The underneath semiconductor is the local anode where a chargemediated nucleophilic substitution reaction takes place, which causes silicon atoms to be etched/removed from the substrate. The metal layer, which is not consumed during the process, simply sinks down while the uncovered parts of the substrate form the tips of the NWs. Indeed, no consummation occurs when gold is used, while other metals are partly dissolved in many instances. Because the fabrication step occurs in liquid ambient a final drying step is inherently involved. Under certain conditions of 1) high NW density and 2) high aspectratio of NWs, the surface tension between the residual fluid film and the NWs could induce a selfassembly [16,17] (see Figure 1). The process of the assembly of NWs induced by elastocapillary forces is complex. There are many factors that influence the assembly such as periodicity, height, cross section, and tensile strength of the NWs as well as evaporation rate and the surface tension of the fluid. Elastocapillary selfassembly of NWs is an extensively investigated versatile and scalable method to design complex and robust surface nanoarchitectures [18]. For tuning and selectivity of the design of NW assemblies other approaches should be considered [19].
In hierarchical nanoarchitectures generated by NWs or other elemental nanobjects the selfassembly/selforganization mechanisms are pivotal to generate the assembled structures. In fact, the direct fabrication of such structures by microfabrication or even nanofabrication approaches would be challenging or impossible, given the nanosized dimensions of the basic units. Because the spatial layout is selfdriven in contrast to a heterodirected placement, asymmetric interaction potentials and entropic forces can lead to different aggregation schemes from place to place and across the scales of the generated structures. The control over the spatial arrangement of the assembled elements is a key issue in nanoarchitectonics, because the emerging functionalities of the whole are linked to its geometry. As a fundamental step towards the full command of the nanoarchitectonic design it is thus necessary to characterize the selfgenerated morphologies, in order to be able to discriminate among them and possibly to relate them to growth procedures from one side and to physical properties from another side. In this regard, fractal analysis [20] is an analytical framework fit for the purpose. Indeed, selfassembled patterns derived from aggregative processes, which are omnipresent in nature, have been characterized by their fractal dimension [2123] that contains information about their geometrical structure at multiple scales. However, sometimes the richness of the organization of shape is such that it is impossible to describe it by just one scaling law. In this latter case a shift to multifractal analysis is necessary.
In the present work we show the results of multifractal analysis of nanoarchitectured surfaces of MACE Si NWs. The spontaneous arrangements of the NWs were investigated by using atomic force microscopy (AFM). Among the scanning probe techniques, AFM shows a peculiar capability to quantitatively characterize features with nanoscaled dimensions. To gain insight over the emergence of the organized nanoarchitectures we applied multifractal analysis to the AFM images. We have found that a single fractal dimension is not sufficient to describe the complex geometries of the NW systems. By examining the results of the multifractal analysis we have been able to highlight differences between the generated spatial patterns that we have correlated to the different growth conditions.
A Brief Survey of Multifractal Analysis
Let the fractal object F be a subset of the ddimensional Euclidean space , which is the physical support of F, and be covered by a ddimensional grid of length scale ε. The boxcounting (BC) fractal dimension, or capacity, D_{BC} is defined as
where N(ε) is the number of grid elements that overlap with F. Usually, D_{BC} is determined through the least squares linear fit of ln N(ε) as a function of ε. It is noteworthy to observe that Equation 1 represents a scaling rule that compares how much the detail (quantified by N) in a given pattern changes with the scale (ε), and it shows that for a (mono)fractal a single fractal dimension is able to characterize it across all the length scales [20]. For the deterministic fractals, which are mathematically constructed objects, the scale invariance holds for all scales. Wellknown examples are Cantor set and Koch’s curve [20]. Instead, natural objects and phenomena are intrinsically finite and their fractality, if any, can be determined only within a specific regime of length scales. These structures are called random fractals.
When the fractal analysis is applied to investigate shapes of natural objects, this is performed by analyzing their images. Thus, it is necessary to reframe the above concepts within the field of image analysis. The simplest type of digital image is a binary image, that is a squared (for the sake of simplicity) S × S discrete matrix M of pixels where each pixel can have black or white colour. In this case the fractal object will correspond to the set of the pixels of a given colour, for instance black, while all pixels corresponding to the other colour will be disregarded. To evaluate the dimension of D_{BC}, first a series of grids with different length scales will be overlaid to the image. Each grid is composed of s_{ε} × s_{ε} not overlapping boxes G_{ε}(i,j), the size of which is the grid length scale ε, such that . Then it is useful to introduce a local measure μ_{ε} (i,j), which amounts to the number of pixels belonging to the fractal object and contained in the box G_{ε}(i,j). It is evident that, when a single global exponent characterizes a fractal object, the measure is uniform and . However, in case of multifractal objects the above does not apply anymore and the measure μ_{ε} varies at different locations.
The quantitative description of multifractality can be performed in differrent manners. One approach passes through the calculation of the LipshitzHölder exponent α, which gives account of the pointwise singularity of the object, and its distribution f(α), known as the multifractal spectrum. One method to determine f(α) is the following [24]. The probability distribution of μ_{ε} is introduced as
from which a oneparameter family of normalized measures is constructed:
The parameter q works like a magnifying glass, enhancing 1) the regions of the fractal object with the lowest values of P_{ε}(i,j) for q < 1 and 2) the regions with the highest values of P_{ε}(i,j) for q > 1. The fractal dimension of the support of μ(q) is
and the average value of the singularity strength
with respect to μ(q) is
The mass exponent τ(q) is defined as
given that
is the qth power moment sum. τ(q) is linked to the multifractal spectrum by the Legendre transformation
It is also connected to the generalized fractal dimensions [2527]
because τ(q) = (q − 1)D(q). Thus, an alternative way to determine the multifractal spectrum is to calculate D(q) from the above equation and to substitute it in Equation 7.
Results and Discussion
Elastocapillary selfassembly in MACE Si NWs
Figure 2 shows typical AFM images of the MACE Si NWs investigated in the present work. The procedure to grow the MACE samples is carried out as follows: (100)oriented Si wafers are the substrates. As a first step, the native oxide is removed from their surfaces by UVoxidizing (2 min) and then dipping them (5 min) in 5% HF. Subsequently, 2 nm thick Au layers are deposited on the cleaned surfaces by electron beam evaporation. These gold films do not coat the substrates uniformly. The uncovered parts of the Si surfaces become the seeds of the NWs in the subsequent etching step. For the etching step, the substrates are immersed in an aqueous solution of HF (5 M) and H_{2}O_{2} (0.44 M). Two types of MACE Si NWs have been synthesised, with differences in 1) the doping of the source substrates and 2) the etching time. Longer etching times yield longer NWs. For one sample (labelled from now on SiNW1; Figure 2a) the source substrate was Pdoped with a doping density of 10^{15} cm^{−3} and a NW length of 5 μm. For the other sample (labelled SiNW2; Figure 2b) the source substrate was Asdoped with a doping density of about 10^{18} cm^{−3} and a NW length of 1.3 μm.
Figure 3a and Figure 3b are the masked images obtained applying a height threshold on the AFM measurements of Figure 2. The threshold divides the surface into two regions: the tips of NWs (black colour) and the remainder of the sample (white colour). The NW tips appear clustered in both samples, creating a very complex architecture over the surfaces. It is noteworthy to observe that this clustering has not been purposefully induced by design of the locations of NWs, but it is a spontaneous assembly occurring during the NW growth. Recently, a direct observation of MACE NWs bending and sticking together during the drying step has been reported [28]. The authors have found experimentally that the bending/bundling of NWs depended on their aspect ratio. It occurred for aspect ratios greater than 1:10, while it was not observed for an aspect ratio of about 1:5. Actually, many more factors have impact on the selfassembly of arrays of nano and microstructures with high aspect ratios, when a liquid is evaporated off the surface [29,30]. Considering two adjacent NWs, first 1) the capillary force between them should be able to overcome the elastic force moving them back to the original straight position, in order to bring them into contact during the evaporation of the liquid. Then a stable bundle will occur if 2) the adhesion force between the surfaces of the two NWs is larger than the elastic force. In order to theoretically confirm the selfassembly that occurred in both samples SiNW1 and SiNW2, we availed of literature [29,31] to model steps 1) and 2). Given a pillarlike structure of height h, clamped at one end, such that the other end is deflected by a length w, the magnitude of the elastic force acting on it can be calculated from the Euler–Bernoulli elementary beam theory [32], and it is given by
where E is the Young modulus of the pillar and I is its area moment of inertia. The magnitude of the capillary force F_{C} between two cylindrical pillars when partially immersed in a liquid is [29]
where γ_{liq} is the liquid surface tension, θ is the contact angle between the liquid and the surface of the pillar, r is the radius of the pillar and 2x is the interdistance between the axes of the two pillars (see Figure 1). We have availed of the above Equations 9 and 10 to estimate the magnitudes of the elastic and capillary forces in our case. To this aim we have determined average values of 4.5 nm and of 7.5 nm for r and x, respectively. Considering the data reported in literature, we have considered not a single value but a range of values of a) E = 80–120 GPa [33] and of b) γ_{HF} from 0.5 mN/m [34] to 10.2 mN/m [35]. A value of θ = 70° has been found for silicon surfaces and HF [36]. The range of values for the capillary force has thus been estimated to be 3–65 pN, which is orders of magnitude greater than the possible values of the elastic force, ranging from 0.9 to 1.4 fN for SiNW1, and from 53 to 80 fN for SiNW2. This confirms that under the growth conditions described here, F_{C} is greater than F_{el} and can bring the NWs into contact.
In order to find out whether the formed NW bundles are stable, we have calculated the critical aspect ratio [31] of the NWs of samples SiNW1 and SiNW2. The critical aspect ratio is a threshold value for the stability of the bundles. NWs with aspect ratios larger than the critical one will remain attached after bending towards each other and getting in contact. Given two collapsed pillars, such that the length of the noncontact portion is L (Figure 1b), a critical value of L can be estimated from [31]
where the term on the lefthand side is the decrease in strain energy if the noncontact region is increased by dL, and the term on the righthand side is the energy required to separate the surfaces of the two pillars by dL. γ_{sup} is the surface energy of the pillar, c_{0} is the contact width at equilibrium of the two pillars under no external force, and U_{c} is the stored elastic energy normalized to the contact length due to the deformation near the contact region. The values of c_{0} and U_{c} are given by [31]
and
given that E* = E/(1 − ν^{2}) where ν is the Poisson’s ratio. By rearranging Equation 11 it is possible to derive the critical aspect ratio as
since I = πr^{4}/4 for a cylindrical pillar. Again, a range of values for γ_{sup} from 1 to 2.2 J/m^{2} has been used according to literature [38], and a Poisson’s ratio of 0.22. The values of AR_{crit} thus obtained range from 2.2 to 1.5, while the aspect ratios of the NWs are of 555 for sample SiNW1 and of 144 for sample SiNW2. This confirms theoretically the stability of the observed NW clusters. The process of selfassembly yielded the hierarchically structured MACE NW surfaces as shown in Figure 2. The complex spatial design thus achieved has been investigated by means of multifractal analysis.
Multifractal analysis applied to MACE NWs
In a previous section we gave a brief survey of the computational procedures involved into performing fractal and multifractal analysis. In both kinds of analysis, after the preliminary step of overlaying a grid of length scale ε to the image, it is crucial to establish a rule to assign a value to the local measure μ_{ε}(i,j) over each box of the grid. However, the recipe that has been given, μ_{ε}(i,j) ↔ number of pixels belonging to the fractal object and contained in the box (i,j) is meaningful only for binary images. Indeed, fractal objects that can be described by binary images are, for example, 2D contours or the 2D correspondent filled patterns (such as the regions of homogeneous colour of Figure 3). These fractals are clearly just a subset of the possible random fractal phenomena or structures that can be met in nature, where the complex features of real 3D morphologies cannot be rendered by binary images. In fact, the AFM images of Figure 2 are RGB images, where a colour scale connects the image colours to height values. It is noteworthy to observe that, once a suitable mapping is established to compute the local measure μ_{ε} for RGB or greyscale images, the steps already described to calculate the fractal dimension, the multifractal spectrum or the generalized dimensions would be the same. In the present work we have chosen to use greyscaled versions of the AFM images to perform the multifractal analysis, which has been implemented by means of the FracLac plugin [39] of the image analysis software ImageJ [40]. In fact, in the greyscaled AFM images the interval of heights actually measured is mapped onto greylevel values ranging from 0 (black) to 255 (white), which corresponds to a simple rescaling. The flow chart of the followed procedure has been schematically illustrated in Figure 4a,b. To calculate the local measure μ_{ε} for the greylevel images we availed of the differential box counting method: μ_{ε}(i,j) = Δ_{ε}I(i,j) + 1, given that Δ_{ε}I(i,j) = I_{max} − I_{min} is the difference in pixel intensities (greylevel values) over the box (i,j).
Figure 5 shows the D_{q} curves of different sampled areas of the surfaces of samples SiNW1 (a) and SiNW2 (b). First of all, these graphical spectra of D_{q} show that the nanoarchitectured surfaces generated by the selfassembly of NWs are indeed multifractal. Monofractals or objects that are not fractals tend to have flatter D_{q} curves than multifractals. Ideally, a D_{q} curve is flat for a monofractal because [25]. With reference to Equation 8, it should be remembered that the generalized dimension is linked to the probability distribution P_{ε}(i,j). The largeness of an element of such a distribution is directly related to the correspondent largeness of μ_{ε}(i,j) (see Equation 2). Thus, the parameter q is a kind of a resolution parameter that enhances 1) regions corresponding to higher μ_{ε} values for positive values of q, and 2) regions of lower μ_{ε} values for negative values of q. Keeping this in mind, and taking into account how we have defined above the local measure μ_{ε}, we observe that for sample SiNW1 the D_{q} values for each q are lower than the ones for sample SiNW2, or they are in the same range. That is, in sample SiNW2 there is a tendency to higher fractality for both areas with larger or smaller ranges of pixel intensities Δ_{ε}I. This is related to the different growth conditions of sample SiNW1 and sample SiNW2.
In Table 1 are reported the D_{0}, D_{1} and D_{2} values of the sampled areas from sample SiNW1 and SiNW2. It can be proved that if q’ > q[25], and in fact we have found that D_{0} > D_{1} > D_{2}, where the difference between these values is an indication of the multifractality of the surfaces of samples SiNW1 and SiNW2. D_{0} is the so called capacity (or box counting) dimension that would coincide with D_{BC} in a monofractal system. In our case we have found values of D_{0} of about 1.84 for most of sampled areas of sample SiNW1, while a value of 1.72 was obtained in one case. In contrast, for sample SiNW2 we obtained values of about 1.73, again with a value of 1.86 obtained in one case. Such differences in capacity dimension for different areas of the same sample may be indicative of a certain degree of inhomogeneity in the final design of the NW patterns. Nevertheless, it is interesting to observe that the range of possible D_{0} values seems to be approximately the same in both samples. It has to be noted that MACE Si NW samples have been characterized by simple fractal analysis in a previous work [37], where a value of D_{BC} of about 1.9 has been found. However, it should be taken into account that in that case the analysis has been performed over binarized SEM images, where the patterns under study have been the regions corresponding to the tops of NWs, similar to the masked regions of Figure 3b,c. Thus, the complex 3D arrangements of the NWs have been completely ignored.
Table 1: Main parameters obtained from the multifractal analysis of the sampled areas of SiNW1 and SiNW2.
SiNW1  

area  D_{0}  D_{1}  D_{2} 
1A  1.7958  1.7854  1.7629 
1B  1.7197  1.7054  1.6878 
1C  1.8528  1.8486  1.8394 
1D  1.8528  1.8453  1.8352 
SiNW2  
area  D_{0}  D_{1}  D_{2} 
2A  1.7197  1.714  1.71 
2B  1.8628  1.8526  1.8435 
2C  1.7378  1.7289  1.7209 
2D  1.7378  1.7296  1.7225 
Finally, in Figure 6 the multifractal spectra f(α) of all the sampled areas of samples SiNW1 (a) and SiNW2 (b) are reported. Again, there is an indication of multifractality in both samples, given that in the case of an ideal monofractal f(α) would reduce to just a single point. In contrast to D_{q}, which represents the various dimensions of the distribution of the ranges of pixel intensity values over the whole imaged area, f(α) is the dimension obtained over different subregions that display the same α. Each curve f(α) shows the characteristic convex shape, peaked at α(0), where f(α(0)) = D_{0}, as can be easily verified by Equations 6–8. The α values to the left of α(0) are associated with positive q values, the ones to the right of α(0) are associated with negative q values. Thus, the presence or the absence of symmetry of f(α) around its peak mirrors the same kind of symmetry/asymmetry in the distribution between regions with a large/small range of pixel intensity values. From Figure 6 it appears that f(α) spectra of the sampled areas of SiNW1 have a tendency to be symmetric. Instead of this, in the case of sample SiNW2 the intervals of α values of subareas with a smaller range of pixel intensity values (right part of the curve) are bigger than the ones for subareas with a larger range of pixel intensity values (left part of the curve). It can be suggested that for sample SiNW2 the multifractality is enhanced in subareas with smaller range of pixel intensity, that is of smaller height variation. In fact, it should be remembered that the pixel intensity is simply a rescaling of the measured height value. It is interesting to note that the property of symmetry/asymmetry of the f(α) spectrum applies to all areas of each sample, appearing to be an intrinsic feature.
Conclusion
In this work we have investigated the spontaneous spatial organization of nanoarchitectures of MACE Si NWs. First, we have applied theoretical models in order to estimate the driving forces leading to the selfassembly. Then we have availed of multifractal analysis to analyze the patterns. Our results confirm that fractal analysis would not be sufficient to capture the whole richness of the selfassembled structures. Differences in growth conditions result in differences in the generalized dimension and the multifractal spectrum. In contrast, when the same quantities are calculated over areas of the same sample the results show coherence, even if a minority of cases departs. This means that the nanoarchitectonic surfaces present locally some degree of inhomogeneity. This also highlights multifractal analysis as a powerful tool to “measure” the design of fractallike nanoarchitectures. Finally, a correlation has been found between the growth conditions and the tendency to multifractality, which is more uniform across the MACE nanoarchitectures of sample SiNW1, while it is accentuated in the region formed by the tips of the nanowires of sample SiNW2.
References

Ariga, K., Ed. Manipulation of Nanoscale Materials; Nanoscience & Nanotechnology Series; Royal Society of Chemistry: Cambridge, United Kingdom, 2012. doi:10.1039/9781849735124
Return to citation in text: [1] 
Ariga, K.; Nishikawa, M.; Mori, T.; Takeya, J.; Shrestha, L. K.; Hill, J. P. Sci. Technol. Adv. Mater. 2019, 20, 51–95. doi:10.1080/14686996.2018.1553108
Return to citation in text: [1] 
Bishop, K. J. M.; Wilmer, C. E.; Soh, S.; Grzybowski, B. A. Small 2009, 5, 1600–1630. doi:10.1002/smll.200900358
Return to citation in text: [1] 
Cui, Y.; Duan, X.; Hu, J.; Lieber, C. M. J. Phys. Chem. B 2000, 104, 5213–5216. doi:10.1021/jp0009305
Return to citation in text: [1] 
Cui, Y.; Lieber, C. M. Science 2001, 291, 851–853. doi:10.1126/science.291.5505.851
Return to citation in text: [1] 
Hasan, M.; Huq, M. F.; Mahmood, Z. H. SpringerPlus 2013, 2, 151. doi:10.1186/219318012151
Return to citation in text: [1] 
Cavallini, A.; Carapezzi, S.; Castaldini, A.; Irrera, A. Phys. B (Amsterdam, Neth.) 2014, 439, 41. doi:10.1016/j.physb.2013.11.021
Return to citation in text: [1] 
Hobbs, R. G.; Petkov, N.; Holmes, J. D. Chem. Mater. 2012, 24, 1975–1991. doi:10.1021/cm300570n
Return to citation in text: [1] 
He, B.; Yang, Y.; Yuen, M. F.; Chen, X. F.; Lee, C. S.; Zhang, W. J. Nano Today 2013, 8, 265–289. doi:10.1016/j.nantod.2013.04.008
Return to citation in text: [1] 
Carapezzi, S.; Castaldini, A.; Mancarella, F.; Poggi, A.; Cavallini, A. ACS Appl. Mater. Interfaces 2016, 8, 10443–10450. doi:10.1021/acsami.6b00600
Return to citation in text: [1] 
Madou, M. J. Wet Chemical Etching and Wet Bulk MicromachiningPools as Tools. Fundamentals of Microfabrication and Nanotechnology, 3rd ed.; Taylor & Francis Group: Boca Raton, FL, U.S.A., 2011; pp 215–318.
Return to citation in text: [1] 
Huang, Z.; Geyer, N.; Werner, P.; de Boor, J.; Gösele, U. Adv. Mater. (Weinheim, Ger.) 2011, 23, 285–308. doi:10.1002/adma.201001784
Return to citation in text: [1] 
Li, X. Curr. Opin. Solid State Mater. Sci. 2012, 16, 71–81. doi:10.1016/j.cossms.2011.11.002
Return to citation in text: [1] 
Han, H.; Huang, Z.; Lee, W. Nano Today 2014, 9, 271–304. doi:10.1016/j.nantod.2014.04.013
Return to citation in text: [1] 
Li, M.; Li, Y.; Liu, W.; Yue, L.; Li, R.; Luo, Y.; Trevor, M.; Jiang, B.; Bai, F.; Fu, P.; Zhao, Y.; Shen, C.; Mbengue, J. M. Mater. Res. Bull. 2016, 76, 436–449. doi:10.1016/j.materresbull.2016.01.006
Return to citation in text: [1] 
Bai, F.; Li, M.; Song, D.; Yu, H.; Jiang, B.; Li, Y. J. Solid State Chem. 2012, 196, 596–600. doi:10.1016/j.jssc.2012.07.029
Return to citation in text: [1] 
Hill, J. J.; Haller, K.; Gelfand, B.; Ziegler, K. J. ACS Appl. Mater. Interfaces 2010, 2, 1992–1998. doi:10.1021/am100290z
Return to citation in text: [1] 
De Volder, M.; Hart, A. J. Angew. Chem., Int. Ed. 2013, 52, 2412–2425. doi:10.1002/anie.201205944
Return to citation in text: [1] 
Carapezzi, S.; Priante, G.; Grillo, V.; Montès, L.; Rubini, S.; Cavallini, A. ACS Nano 2014, 8, 8932–8941. doi:10.1021/nn503629d
Return to citation in text: [1] 
Mandelbrot, B. B. The fractal geometry of nature; W. H. Freeman and Co., 1982.
Return to citation in text: [1] [2] [3] 
Carrillo, J. L.; Mendoza, M. E.; Donado, F. J. Stat. Mech.: Theory Exp. 2005, P06001. doi:10.1088/17425468/2005/06/p06001
Return to citation in text: [1] 
GonzálezGutiérrez, J.; CarrilloEstrada, J. L.; RuizSuárez, J. C. J. Stat. Mech.: Theory Exp. 2013, P12015. doi:10.1088/17425468/2013/12/p12015
Return to citation in text: [1] 
GonzálezGutiérrez, J.; CarrilloEstrada, J. L.; RuizSuárez, J. C. J. Phys.: Conf. Ser. 2013, 475, 012003. doi:10.1088/17426596/475/1/012003
Return to citation in text: [1] 
Chhabra, A.; Jensen, R. V. Phys. Rev. Lett. 1989, 62, 1327–1330. doi:10.1103/physrevlett.62.1327
Return to citation in text: [1] 
Hentschel, H. G. E.; Procaccia, I. Phys. D (Amsterdam, Neth.) 1983, 8, 435–444. doi:10.1016/01672789(83)90235x
Return to citation in text: [1] [2] [3] 
Halsey, T. C.; Meakin, P.; Procaccia, I. Phys. Rev. Lett. 1986, 56, 854–857. doi:10.1103/physrevlett.56.854
Return to citation in text: [1] 
Halsey, T. C.; Jensen, M. H.; Kadanoff, L. P.; Procaccia, I.; Shraiman, B. I. Phys. Rev. A 1986, 33, 1141–1151. doi:10.1103/physreva.33.1141
Return to citation in text: [1] 
Kara, S. A.; Keffous, A.; Giovannozzi, A. M.; Rossi, A. M.; Cara, E.; D'Ortenzi, L.; Sparnacci, K.; Boarino, L.; Gabouze, N.; Soukane, S. RSC Adv. 2016, 6, 93649–93659. doi:10.1039/c6ra20323j
Return to citation in text: [1] 
Chandra, D.; Yang, S. Langmuir 2009, 25, 10430–10434. doi:10.1021/la901722g
Return to citation in text: [1] [2] [3] 
Chandra, D.; Yang, S. Acc. Chem. Res. 2010, 43, 1080–1091. doi:10.1021/ar100001a
Return to citation in text: [1] 
Glassmaker, N. J.; Jagota, A.; Hui, C.Y.; Kim, J. J. R. Soc., Interface 2004, 1, 23–33. doi:10.1098/rsif.2004.0004
Return to citation in text: [1] [2] [3] [4] 
Hibbeler, R. C. Mechanics of materials, 3rd ed.; PrenticeHall: Englewood Cliffs, NJ, U.S.A., 1997.
Return to citation in text: [1] 
Zhu, Y.; Xu, F.; Qin, Q.; Fung, W. Y.; Lu, W. Nano Lett. 2009, 9, 3934–3939. doi:10.1021/nl902132w
Return to citation in text: [1] 
dos Santos, A. P.; Levin, Y. J. Chem. Phys. 2010, 133, 154107. doi:10.1063/1.3505314
Return to citation in text: [1] 
Hydrofluoric acid  HF  PubChem. https://pubchem.ncbi.nlm.nih.gov/compound/14917#section=HeatofVaporization&fullscreen=true (accessed Oct 10, 2019).
Return to citation in text: [1] 
Hermansson, K.; Lindberg, U.; Hok, B.; Palmskog, G. Wetting properties of silicon surfaces. In TRANSDUCERS 91: 1991 International Conference on SolidState Sensors and Actuators. Digest of Technical Papers, IEEE Publishing: Piscataway, U.S.A., 1991; pp 193–196.
Return to citation in text: [1] 
Fazio, B.; Artoni, P.; Antonia Iatì, M.; D'Andrea, C.; Lo Faro, M. J.; Del Sorbo, S.; Pirotta, S.; Giuseppe Gucciardi, P.; Musumeci, P.; Salvatore Vasi, C.; Saija, R.; Galli, M.; Priolo, F.; Irrera, A. Light: Sci. Appl. 2016, 5, e16062. doi:10.1038/lsa.2016.62
Return to citation in text: [1] [2] 
Tran, R.; Xu, Z.; Radhakrishnan, B.; Winston, D.; Sun, W.; Persson, K. A.; Ong, S. P. Sci. Data 2016, 3, 160080. doi:10.1038/sdata.2016.80
Return to citation in text: [1] 
FracLac for ImageJ. https://rsb.info.nih.gov/ij/plugins/fraclac/FLHelp/Introduction.htm (accessed Oct 10, 2019).
Return to citation in text: [1] 
ImageJ. https://imagej.nih.gov/ij/ (accessed Oct 10, 2019).
Return to citation in text: [1]
31.  Glassmaker, N. J.; Jagota, A.; Hui, C.Y.; Kim, J. J. R. Soc., Interface 2004, 1, 23–33. doi:10.1098/rsif.2004.0004 
31.  Glassmaker, N. J.; Jagota, A.; Hui, C.Y.; Kim, J. J. R. Soc., Interface 2004, 1, 23–33. doi:10.1098/rsif.2004.0004 
38.  Tran, R.; Xu, Z.; Radhakrishnan, B.; Winston, D.; Sun, W.; Persson, K. A.; Ong, S. P. Sci. Data 2016, 3, 160080. doi:10.1038/sdata.2016.80 
1.  Ariga, K., Ed. Manipulation of Nanoscale Materials; Nanoscience & Nanotechnology Series; Royal Society of Chemistry: Cambridge, United Kingdom, 2012. doi:10.1039/9781849735124 
2.  Ariga, K.; Nishikawa, M.; Mori, T.; Takeya, J.; Shrestha, L. K.; Hill, J. P. Sci. Technol. Adv. Mater. 2019, 20, 51–95. doi:10.1080/14686996.2018.1553108 
9.  He, B.; Yang, Y.; Yuen, M. F.; Chen, X. F.; Lee, C. S.; Zhang, W. J. Nano Today 2013, 8, 265–289. doi:10.1016/j.nantod.2013.04.008 
10.  Carapezzi, S.; Castaldini, A.; Mancarella, F.; Poggi, A.; Cavallini, A. ACS Appl. Mater. Interfaces 2016, 8, 10443–10450. doi:10.1021/acsami.6b00600 
24.  Chhabra, A.; Jensen, R. V. Phys. Rev. Lett. 1989, 62, 1327–1330. doi:10.1103/physrevlett.62.1327 
8.  Hobbs, R. G.; Petkov, N.; Holmes, J. D. Chem. Mater. 2012, 24, 1975–1991. doi:10.1021/cm300570n 
25.  Hentschel, H. G. E.; Procaccia, I. Phys. D (Amsterdam, Neth.) 1983, 8, 435–444. doi:10.1016/01672789(83)90235x 
26.  Halsey, T. C.; Meakin, P.; Procaccia, I. Phys. Rev. Lett. 1986, 56, 854–857. doi:10.1103/physrevlett.56.854 
27.  Halsey, T. C.; Jensen, M. H.; Kadanoff, L. P.; Procaccia, I.; Shraiman, B. I. Phys. Rev. A 1986, 33, 1141–1151. doi:10.1103/physreva.33.1141 
4.  Cui, Y.; Duan, X.; Hu, J.; Lieber, C. M. J. Phys. Chem. B 2000, 104, 5213–5216. doi:10.1021/jp0009305 
5.  Cui, Y.; Lieber, C. M. Science 2001, 291, 851–853. doi:10.1126/science.291.5505.851 
6.  Hasan, M.; Huq, M. F.; Mahmood, Z. H. SpringerPlus 2013, 2, 151. doi:10.1186/219318012151 
7.  Cavallini, A.; Carapezzi, S.; Castaldini, A.; Irrera, A. Phys. B (Amsterdam, Neth.) 2014, 439, 41. doi:10.1016/j.physb.2013.11.021 
20.  Mandelbrot, B. B. The fractal geometry of nature; W. H. Freeman and Co., 1982. 
37.  Fazio, B.; Artoni, P.; Antonia Iatì, M.; D'Andrea, C.; Lo Faro, M. J.; Del Sorbo, S.; Pirotta, S.; Giuseppe Gucciardi, P.; Musumeci, P.; Salvatore Vasi, C.; Saija, R.; Galli, M.; Priolo, F.; Irrera, A. Light: Sci. Appl. 2016, 5, e16062. doi:10.1038/lsa.2016.62 
3.  Bishop, K. J. M.; Wilmer, C. E.; Soh, S.; Grzybowski, B. A. Small 2009, 5, 1600–1630. doi:10.1002/smll.200900358 
20.  Mandelbrot, B. B. The fractal geometry of nature; W. H. Freeman and Co., 1982. 
18.  De Volder, M.; Hart, A. J. Angew. Chem., Int. Ed. 2013, 52, 2412–2425. doi:10.1002/anie.201205944 
20.  Mandelbrot, B. B. The fractal geometry of nature; W. H. Freeman and Co., 1982. 
25.  Hentschel, H. G. E.; Procaccia, I. Phys. D (Amsterdam, Neth.) 1983, 8, 435–444. doi:10.1016/01672789(83)90235x 
16.  Bai, F.; Li, M.; Song, D.; Yu, H.; Jiang, B.; Li, Y. J. Solid State Chem. 2012, 196, 596–600. doi:10.1016/j.jssc.2012.07.029 
17.  Hill, J. J.; Haller, K.; Gelfand, B.; Ziegler, K. J. ACS Appl. Mater. Interfaces 2010, 2, 1992–1998. doi:10.1021/am100290z 
21.  Carrillo, J. L.; Mendoza, M. E.; Donado, F. J. Stat. Mech.: Theory Exp. 2005, P06001. doi:10.1088/17425468/2005/06/p06001 
22.  GonzálezGutiérrez, J.; CarrilloEstrada, J. L.; RuizSuárez, J. C. J. Stat. Mech.: Theory Exp. 2013, P12015. doi:10.1088/17425468/2013/12/p12015 
23.  GonzálezGutiérrez, J.; CarrilloEstrada, J. L.; RuizSuárez, J. C. J. Phys.: Conf. Ser. 2013, 475, 012003. doi:10.1088/17426596/475/1/012003 
25.  Hentschel, H. G. E.; Procaccia, I. Phys. D (Amsterdam, Neth.) 1983, 8, 435–444. doi:10.1016/01672789(83)90235x 
12.  Huang, Z.; Geyer, N.; Werner, P.; de Boor, J.; Gösele, U. Adv. Mater. (Weinheim, Ger.) 2011, 23, 285–308. doi:10.1002/adma.201001784 
13.  Li, X. Curr. Opin. Solid State Mater. Sci. 2012, 16, 71–81. doi:10.1016/j.cossms.2011.11.002 
14.  Han, H.; Huang, Z.; Lee, W. Nano Today 2014, 9, 271–304. doi:10.1016/j.nantod.2014.04.013 
15.  Li, M.; Li, Y.; Liu, W.; Yue, L.; Li, R.; Luo, Y.; Trevor, M.; Jiang, B.; Bai, F.; Fu, P.; Zhao, Y.; Shen, C.; Mbengue, J. M. Mater. Res. Bull. 2016, 76, 436–449. doi:10.1016/j.materresbull.2016.01.006 
39.  FracLac for ImageJ. https://rsb.info.nih.gov/ij/plugins/fraclac/FLHelp/Introduction.htm (accessed Oct 10, 2019). 
11.  Madou, M. J. Wet Chemical Etching and Wet Bulk MicromachiningPools as Tools. Fundamentals of Microfabrication and Nanotechnology, 3rd ed.; Taylor & Francis Group: Boca Raton, FL, U.S.A., 2011; pp 215–318. 
19.  Carapezzi, S.; Priante, G.; Grillo, V.; Montès, L.; Rubini, S.; Cavallini, A. ACS Nano 2014, 8, 8932–8941. doi:10.1021/nn503629d 
29.  Chandra, D.; Yang, S. Langmuir 2009, 25, 10430–10434. doi:10.1021/la901722g 
31.  Glassmaker, N. J.; Jagota, A.; Hui, C.Y.; Kim, J. J. R. Soc., Interface 2004, 1, 23–33. doi:10.1098/rsif.2004.0004 
28.  Kara, S. A.; Keffous, A.; Giovannozzi, A. M.; Rossi, A. M.; Cara, E.; D'Ortenzi, L.; Sparnacci, K.; Boarino, L.; Gabouze, N.; Soukane, S. RSC Adv. 2016, 6, 93649–93659. doi:10.1039/c6ra20323j 
29.  Chandra, D.; Yang, S. Langmuir 2009, 25, 10430–10434. doi:10.1021/la901722g 
30.  Chandra, D.; Yang, S. Acc. Chem. Res. 2010, 43, 1080–1091. doi:10.1021/ar100001a 
37.  Fazio, B.; Artoni, P.; Antonia Iatì, M.; D'Andrea, C.; Lo Faro, M. J.; Del Sorbo, S.; Pirotta, S.; Giuseppe Gucciardi, P.; Musumeci, P.; Salvatore Vasi, C.; Saija, R.; Galli, M.; Priolo, F.; Irrera, A. Light: Sci. Appl. 2016, 5, e16062. doi:10.1038/lsa.2016.62 
31.  Glassmaker, N. J.; Jagota, A.; Hui, C.Y.; Kim, J. J. R. Soc., Interface 2004, 1, 23–33. doi:10.1098/rsif.2004.0004 
35.  Hydrofluoric acid  HF  PubChem. https://pubchem.ncbi.nlm.nih.gov/compound/14917#section=HeatofVaporization&fullscreen=true (accessed Oct 10, 2019). 
36.  Hermansson, K.; Lindberg, U.; Hok, B.; Palmskog, G. Wetting properties of silicon surfaces. In TRANSDUCERS 91: 1991 International Conference on SolidState Sensors and Actuators. Digest of Technical Papers, IEEE Publishing: Piscataway, U.S.A., 1991; pp 193–196. 
33.  Zhu, Y.; Xu, F.; Qin, Q.; Fung, W. Y.; Lu, W. Nano Lett. 2009, 9, 3934–3939. doi:10.1021/nl902132w 
34.  dos Santos, A. P.; Levin, Y. J. Chem. Phys. 2010, 133, 154107. doi:10.1063/1.3505314 
32.  Hibbeler, R. C. Mechanics of materials, 3rd ed.; PrenticeHall: Englewood Cliffs, NJ, U.S.A., 1997. 
© 2019 Carapezzi and Cavallini; licensee BeilsteinInstitut.
This is an Open Access article under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0). Please note that the reuse, redistribution and reproduction in particular requires that the authors and source are credited.
The license is subject to the Beilstein Journal of Nanotechnology terms and conditions: (https://www.beilsteinjournals.org/bjnano)