3.1. Structural Properties of Thin Nanoporous Films
In this section, we analyze the nanostructure of the films itself. In the next section, we investigate how they cover the substrate having larger pores.
The typical structural properties of the prepared films are demonstrated in Figure 2
. The model of an ideal network structure is shown in Figure 2
a, while the main parameters of it are illustrated in Figure 2
b. The nanopores make a 3D network having a body centered (BCT) tetragonal arrangement of the nodes. The separation between the nodes in the plane parallel to the substrate is described by the parameter a
, while the vertical separation is c
. The pores have the radius R
and the length L
. The TEM images of the film structure are shown in Figure 2
c,d. The shown films differ by the Ge sputtering power, resulting in the formation of longer, more separated pores for the smaller Ge power. The networks are continuous, which is ensured by the preparation method, i.e., the evaporation of Ge from the film during the sample annealing. The insertion of even a very thin (1 nm) layer of pure matrix prevents Ge oxide evaporation, because the network is not continuous in that case. Therefore, the same preparation method is not efficient for Ge quantum dot multilayer films. In such films, Ge quantum dots are fully surrounded by alumina matrix, which prevents the evaporation of Ge. Therefore, these films cannot be used for the creation of nanoporous films. The GISAXS intensity patterns of the films’ structure are shown in Figure 2
e,f. The two characteristic lateral peaks are related to the arrangement of the network nodes and to their regularity. A more regular network produces stronger and narrower Bragg peaks, so the P1D3 film has a better ordering in this case. More details about the GISAXS maps of such structures are given in [43
]. GISAXS is very suitable for the analysis of these structures, as it provides data about nanopore ordering and size properties with excellent statistics.
The GISAXS maps of all the investigated films before and after annealing (i.e., with Ge quantum wires and nanopores, respectively) are given in Figure 3
. The first three films (Figure 3
a–c) differ only by the film thickness (D1–D3), while the last two (Figure 3
d,e) have different Ge sputtering powers (P2, P3), i.e., different pore arrangements and size properties. The side maxima in the GISAXS maps (Bragg spots) are at nearly the same positions for the first three films, showing the homogeneity of the structure, because these films differ only by their thickness. For the last two films, the spots become more elongated and more spaced, showing an increase in the disorder and a decrease in the lattice parameters of the formed 3D pore network.
All the films except the last one have nearly the same position of the Bragg spots before and after annealing (first and second rows in Figure 3
, respectively), showing that during annealing, the Ge left the film, leaving empty space behind without affecting the alumina. The GISAXS method is sensitive to the electron density contrast between the nano-objects and the matrix. The Ge and alumina have a strong difference in electron density, as well as in the empty space (pores) and alumina. Therefore, we see nearly the same GISAXS maps from the Ge nanowires as from the pores of the same shape in the alumina matrix. Only the last film changed the structure during annealing, probably due to the high percentage of Ge that caused a collapse of the structure after leaving the film. This follows clearly from the GISAXS maps of that film (P3D1, shown in Figure 3
c). The GISAXS map of the annealed film (P3D1 annealed) differs significantly from the map of the as-grown film. This fact shows clear evidence that the internal structure changed significantly during annealing. The lateral peaks called Bragg spots are clearly visible in the As-grown film, while they are absent for the annealed one, showing that there is no regular ordering of the pores after annealing.
The GISAXS maps were numerically analyzed using the procedure described in [44
] in order to obtain the values of the structural parameters of the pores and their arrangement. The results of the analysis are given in Table 2
. From the results, we can see that the nanopore radius (R
) is around 0.6 nm for all the films. This is due to the same deposition temperature for all the films, as shown in the case of Ge quantum wires in [46
]. The deposition time (D1–D3) only significantly influences the thickness of the films (D), assuming a constant Ge sputtering power (P1), while the pore length (L
) depends on the Ge sputtering power during the film preparation (P1–P3). The pore length decreases with increasing Ge sputtering power. We want to point out here that the pore length L
refers to the length of the pore between the two adjacent nodes of the BCT lattice (please see Figure 2
b), but the pores are interconnected, so the real pore length is much larger. The pore length L is not correlated to the film thickness. The film grows homogeneously with the same value of the pore length. More information about the production of Ge quantum wire lattices with different structural parameters is given in [46
In summary, the structural properties of the nanopores are determined by the properties of the Ge quantum wire lattices. By tuning the deposition parameters, it is possible to produce networks of nanopores having different radii and lengths.
3.2. Growth of Nanoporous Thin Films on Alumina Substrate with Larger Pores
In this section, we investigate the preparation of nanomembranes using the above-mentioned films. As described in the Methods section, we deposited the films on porous alumina substrates (Whatman®
Anodisc Inorganic Membranes) having larger pores. The used substrate membranes have different pore sizes on their sides as visible from the SEM measurements, shown in Figure 4
. Side 1 has pores with a diameter of 20–40 nm (Figure 4
a), and Side 2 has larger pores with a diameter of about 200 nm (Figure 4
The effect of the film deposition on Side 1 of the membrane is shown in Figure 5
. The SEM images of the film’s surface clearly show the coverage of the pores with increasing film thickness D
. The thinnest film P1D1 shown in Figure 5
a does not cover the pores. The coverage is complete for the middle thickness (40 nm) shown in Figure 5
b. This indicates that the minimal film thickness needed to achieve the full coverage of the substrate holes is found to be nearly equal to the lateral size of the holes. However, features similar to cracks are visible in the film structure. A complete coverage and a smooth film surface are achieved for the film P1D3 having the largest thickness of about 80 nm.
The coverage of the larger pores (Side 2) of the AAO substrate by the films is shown in Figure 6
. The films differing by the deposition time are shown in Figure 6
a,b, and the films differing by Ge sputtering power are shown in Figure 6
c,d. The pores are not fully covered in all cases; however, the pore sizes are strongly reduced for the longest deposition time (P1D3) and for the largest Ge sputtering power (P3D1). Since the reduction of the pore size is proportional to the film thickness, this technique can be applied for tailoring the pore size to desired value.
The effect of the pore coverage is also investigated on a single opening (pore) in the SiNx
membrane; see Figure 7
. The single opening of initial diameter of ~220 nm was investigated by TEM to inspect the overgrowth behavior of the films on different substrates. Figure 7
a shows the opening having a diameter of ~120 nm after the growth of the film P3D1 with the thickness of 42 nm. The non-covered part of the opening is visible as a circular shape in light-gray color. The grown film appears slightly darker, while the rest of the membrane exhibits a dark gray color. The effect is better visible in Figure 7
b where the enlarged part of the opening is shown. The nanopores in the film are visible, and the thickness of the covered part of the opening is about 40 nm. As visible from the images, the film narrowed the opening for the value of its thickness, shrinking the pore down to ~120 nm. The nanoporous structure of the film is shown in Figure 7
c. Results for the reduction of the pore size in the SiNx membrane shown in Figure 7
a–c, are shown in Figure 7
d. The curve shows a negative slope with the value of 2.26, which may be used for the estimation of the pore size reduction using the presented technique.