Creation of unique shapes by coordination of alumina nanopores and carbon nanowalls

Abstract This work presents experimental results on the synthesis of сarbon nanowalls (CNWs) with predefined morphology on the surface of the nanoporous alumina membrane using two different methods, namely radio-frequency plasma-enhanced chemical vapor deposition (RF-PECVD) and radical-injection (RI)-PECVD. Obtained samples were characterized by the methods of scanning electron microscopy (SEM), transmission electron microscopy (TEM), and Raman spectroscopy. From the microstructure analyses of CNWs, it has been observed that there is a time dependence on the reproducibility of membrane morphology by CNWs. At the early stage of nanowalls growth, nanowalls prefer to grow around the edges of nanopores and continue to grow vertically with time. In RF-PECVD, the nanopores’ size begins to shrink drastically and pores are completely covered by secondary flake-like nanowalls after 25 minutes of growth. In the case of CNWs grown using RI-PECVD, nanowalls are more vertical and self-supported. This is because of the rapid and sustained production of hydrogen radicals that prevent the secondary growth of carbon nanowalls. In addition, the influence of pores diameter and membrane thickness on the growth of RI-PECVD CNWs was revealed.


Introduction
Carbon nanowalls (CNWs) are one of the allotropic modifications of carbon and can be described as three-dimensional networks of vertically oriented graphene layers. [1,2] CNWs consist of interconnecting flakes or wall-like aggregations of graphene sheets standing vertically on a substrate. As compared to classic graphene, carbon nanowalls typically have a large specific surface area. At the same time, due to the vertical orientation of small graphene sheets, there is always rather a high level of defectiveness contributed by crystallite edges. [3,4] CNWs can be synthesized on various metallic, semiconducting, and dielectric substrates. [5,6] using different methods. [7][8][9][10][11][12][13] However, it should be noted that most of those methods use excitation or enhancement by any plasma. The versatility of CNWs for various applications was proved by the investigation of their structural, physical, and chemical properties, thus, high specific surface area, mechanical strength, attractive electrical, optical, and other properties [1,2,8,[14][15][16] of CNWs open great opportunities for their practical applications, in particular for the creation of light-emitting diodes, sensors, solar cells, superhydrophobic coatings, supercapacitors, memory devices, etc. [2,5,7,[17][18][19][20][21][22] Despite the wide range of possible applications, control of the synthesis process and final morphology of CNW films is still a complicated task, particularly obtaining of CNWs with the required morphology and properties. [14,23,24] There are already several papers on the control of CNW morphology. For example, the growth and shape of carbon nanowalls can be controlled through the adjustment of carbon atom concentration. [25][26][27] Authors of the work [4] reported that it is possible to control the average distance between the walls by changing the parameters of plasma discharge. Thus, they managed to control the wall separation distance using the system of radical injection RI-PECVD simultaneously changing the input voltage V in from 90 V to 150 V to change the electrostatic power of impulses. It was revealed that the higher voltages lead to a wider distance between the walls.
In work [28] , the authors found out that by changing the rate of introduction of feed gases during the PECVD process, it is possible to change the distance between nanowalls which in turn allows controlling the structural properties such as surface area and crystallinity of CNWs. Authors of papers [24,28] made an unsuccessful attempt to obtain CNWs with predefined morphology, however, using different metal coatings they obtained CNWs with different morphology.
Nanoporous alumina membranes with pre-designed porous structures and pore density make them an ideal template to grow micro-and nano-structures of different materials for various applications such as biosensors, biomedical, and drug delivery applications. [29] This work is dedicated to the synthesis of CNWs on nanoporous alumina membranes as substrate with predefined morphology using plasma-enhanced chemical vapor deposition (PECVD). The nanoporous alumina membranes with different morphology and thickness were obtained by the method of two-step electrochemical anodization. Synthesized carbon nanostructures were studied using Raman spectroscopy, scanning electron microscopy, and transmitting electron microscopy. The dependence of morphology and the height of carbon nanowalls on the synthesis time, pore diameter, and thickness of nanoporous alumina membrane is revealed.

Obtaining nanoporous alumina membranes
Nanoporous aluminum oxide membranes used as substrates were produced by two-stage electrochemical anodization of high purity (99.997%) aluminum foil carried out in 0.4 M aqueous solutions of H 3 PO 4 at 22-25 C and 100-120 V. This process is described in more detail in the works [14,30] . As a result of anodization nanoporous aluminum oxide membranes with a pore diameter of $150-200 nm and a membrane thickness of 3.8 and 10.7 mm were obtained.

Synthesis of CNWs by RF-PECVD and RI-PECVD methods
Synthesis of CNWs on the surface of nanoporous alumina membranes was conducted using two different methods. The first one is RF (radio frequency), the PECVD method, detailed description of the experimental setup and the synthesis process is reported in papers. [15,31] Synthesis parameters in this method are the following: frequency-13.56 MHz, power-11 W, heater temperature-500 j (substrate temperature-460 C), duration of Ar plasma processing-10 min at the flow rate 7 sccm, synthesis time 25-35 min in the flow of Ar/methane gas mixture À 7/0.8 sccm, respectively. The second applied method is RI (radical injection), the PECVD method. Its deposition system has a surface wave plasma (SWP) source with a conventional capacitively coupled plasma (CCP) between parallel plates. Using the SWP source, high-density atomic hydrogen radicals can be generated by microwave and injected into the CCP region. Parameters of the RI-PECVD experiment: SWP frequency and power 2.45 GHz and 400 W, CCP frequency 100 GHz and 100 W, the CH 4 to H 2 flow ratio in this study was kept as 1:2, i.e., 50 and 100 cm 3 /min, respectively, substrate surface temperature 460 C, synthesis time 10 min. More detailed information about this technique is presented in works. [14,32,33]

Materials characterization and methods
Morphology of the obtained CNW films is characterized using a scanning electron microscope (SEM, Hitachi High-Technologies SU8200 and Quanta 3 D 200i FEI Company), whereas a transmission electron microscope (TEM, JEOL JEM À 1400 Plus) and a Raman spectrometer (Solver Spectrum, NT-MDT with the laser wavelength of 473 nm) are used to investigate the structural properties of the samples. The Minkowski functionals analysis was carried out using the Gwyddion 2.55 program. [34,35] 3. Results and discussions Figure 1 shows SEM images of CNWs samples synthesized on the surface of nanoporous alumina membrane using the RF-PECVD method. Before the deposition, the membrane demonstrates a pore diameter of $150 nm ( Figure 1a) and a thickness (or pore depth) of 10 micrometers. After 25 minutes of the synthesis, one can see that CNWs replicate (reproduce) the membrane's morphology (Figure 1b). At the following increase in the synthesis time to 30 min, the density of CNWs on the membrane surface increases (Figure 1c), and at 35 min, pores of membrane are almost completely covered by CNWs (Figure 1d), since long-term synthesis leads to increasing density and height of the resulting CNWs. [15,36] The morphological properties of the obtained samples were also characterized using Minkowski functional. The connectivity of 2-dimensional discrete variants (Euler-Poincar e Characteristic) v is calculated according to the formula [34] : where N is the total number of pixels, C white and C black show the number of continuous sets of white and black pixels, respectively. This function is used for the description (characterization) of morphological features, which can not be defined using classical methods of image analysis. Minkowski functionals are based on the separation of an image into two partstop and bottomconnecting to the threshold value. Here, tops and bottoms are inflection points that give maximal and minimal values. Currently, this method is widely used for the quantitative assessment of morphological characteristics of nanostructured materials, [37][38][39] including carbon nanowalls. [35,40,41] Curves of two-dimensional Minkowski functional connectivity (v) characterizing the initial nanoporous alumina membrane and CNWs synthesized on it are shown in Figure 2. Minkowski connectivity is calculated from the difference between the numbers of top and bottom areas and characterizes topological patterns (fractal nature). This parameter characterizes the property dependence on the interaction of nanostructures in the network such as percolation threshold, conductivity, and others connected with the transfer of gas, heat, electrons, etc. between nanostructures. [42] On the resulting graph ( Figure 2) negative values correspond to the predominance of valleys with minimal values for the highest density of bottom areas with the maximum/minimum values corresponding to the densest peaks/ pits, while positive values of v indicate the prevalence of high-domain (top) units. It should be noted that the shape of the obtained curves is very similar; however, there is a decrease in the amplitude and a shift of the minimum peak depending on the increase in the synthesis time ( Figure 2). The shift and decrease in the minimum value of v depending on the synthesis time indicate a decrease in areas of valleys (continuously connected black pixels) in the morphology of the obtained samples and a tendency to close the pores of the membrane with CNWs films, as shown in Figure 1. At the same time, on the curves after the synthesis process at different times, a uniform distribution of the positive value of v is noticeable, which indicates an increase in the height of the CNWs in a wide range.
For a complete study and comprehension of the synthesized CNWs' features, a Raman spectroscopy study was performed. Figure 3 shows the results of the structural analysis of CNW films obtained by Raman spectroscopy. The Raman spectra of the samples presented in Figure 3a show a typical CNWs spectrum with clear characteristic graphite peaks D, G, D', G' (2 D), and G þ D. [43] Analysis of Raman spectra (Figure 3b) shows the ratios of peaks' intensities -I D /I G , I G / I 2D , I G /I D'. One can see that with increasing synthesis time, there is a decline in the intensity of peaks D and D' that are induced by defects in graphite structure. Moreover, the decrease in I D /I G ratio indicates the increase in few-layer graphene lateral sizes with increasing synthesis time. This is also proved by SEM images as white lines that correspond to the vertically oriented carbon nanowalls getting longer and higher contrast. A clear downward trend in I G /I 2D value for longer synthesis duration can be induced by increasing levels of graphitization and long-range order. The details of the Raman spectra analysis of the CNW films is in the Supplementary material (see Table S1). Figure 4d presents an SEM image of a nanoporous alumina membrane with a pore diameter of $150 nm before  carbon deposition, it can be seen that its morphology has certain oriented line (scratch-like, Figure S3, Supplementary material). It should be noted that the surface of the synthesized CNW film also demonstrates such lines (Figures 4e  and S4, Supplementary material). The height of the CNWs on the membranes with a thickness of 3.8 and 10.5 micrometers is 671 and 334 nm, respectively ( Figure S5, Supplementary material). The observing difference in the height of the obtained material (CNWs) can be explained by the initial thickness of the membrane (substrate), a more detailed description is reported in the paper. [14] Notably, both the synthesis methods RF-PECVD ( Figure 4) and RI-PECVD (Figure 1) lead to the formation of CNWs that replicate the morphology of the membrane. Raman analysis of CNW obtained by RI-PECVD is shown in Figure 4f. It can be seen that the spectrum is typical for CNW and similar to that of CNW obtained by the RF-PECVD method. A significant difference was found in the I D /I G peak ratios, which are 2.5 and 2.1 for the membrane thicknesses of 3.8 and 10.5 microns, respectively. The details of the Raman spectra analysis of the CNW films are in the Supplementary material (see Table S1). This indicates the decrease of defects in the synthesized structures with increasing substrate thickness. Two-dimensional Minkowski connectivity functionals of carbon nanowalls synthesized by the RI-PECVD method are shown in ( Figure S6, Supplementary material), as can be seen, like in the RF-PECVD method, the intensity of the minimum decreases, and the shift of this peak toward value 1 is observed, this phenomenon has a characteristic correlation with the synthesis method. Figure 5 shows the results of the TEM analysis of the obtained CNWs synthesized by the RI-PECVD method. Sample preparation for TEM measurement is described in detail in the supporting information. Figure 5a shows a top view of the CNW film. The darker areas correspond to vertically oriented CNWs, while the lighter areas correspond to a continuous CNW film oriented horizontally. As seen from a side-view TEM image (Figure 5b) of the CNW film, walls have a preferential vertical orientation. A high-resolution image of the coiled edge of the CNW film (see Figure 5c-f) shows that the walls consist of several layers of graphene  sheets with an average spacing between layers of $0.345 nm. These data match with the d-spacing values for the CNWs reported by other groups. [43] However, from Figure 5c, one can see that there are some areas without any noticeable deposited material, suggesting that these areas correspond to the location of pores in the alumina membrane. Figures 5e  and 5f show that the edge of some CNWs is coiled and arcshaped, which is also in good agreement with the results of other works. [44] Thus, in this work, CNWs with the predefined morphology of a nanoporous aluminum oxide membrane were obtained using two different methods -RF-PECVD and RI-PECVD. According to the obtained results, at the initial stage of the synthesis process, CNWs start to grow on the areas between the membrane's pores; further increasing synthesis time in case of RF-PECVD and saturation with carbon radicals in case of RI-PECVD causes the increase in CNWs density and the membrane pores become covered with CNW film. The effect of synthesis time on the morphology of the CNWs was studied using the RF-PECVD synthesis method. It was found that at 25 min of synthesis, CNWs are almost completely replicated the membrane morphology. At a longer synthesis time, carbon nanowalls keep growing more densely and eventually fill the membrane's pores forming a chaotic and labyrinthlike morphology.

Conclusion
In conclusion, we have demonstrated that well-defined morphology on the alumina membrane was successfully synthesized using the RIPECVD method, which requires a relatively low synthesis temperature (460 C) and less growing time (10 min). In addition, the influence of pores diameter and membrane thickness on the growth of RI-PECVD CNWs was revealed. The obtained Raman spectra of the samples are typical for CNWs, while in the case of RF-PECVD synthesis, there is a definite tendency of improving the quality of CNW with an increase in the synthesis time. The observed results indicate that it is possible to obtain CNWs with a predefined morphology. Therefore, the obtained experimental results can be used to control the morphology of CNWs, which is of great importance for further practical applications of carbon nanostructures for creating various sensors (gas, pressure, light, etc.) and supercapacitors.

Disclosure statement
No conflict of interest has been reported by the authors.

Data availability statement
All data generated or analyzed during this study are included in this published article and its supplementary information files.