In Figure , we give the top section SEM images of sample A, where the silver nanoparticles cover the top surface of the SiNWs. The silver film formed on wafer A is not uniform and not compact. The nonhomogeneity of this film can be attributed to the surface status. It was reported in several works [5
] that the dendrites formation is related to the electrons’ exchange between silicon atoms and silver ions. In Figure a, we observe in some regions silver nanoparticles on top of SiNWs, where the image in Figure b is the magnification of this zone. In other regions, the silver forms a filament-like structure. The growth mechanism of silver dendrites is still unclear; the diffusion-limited aggregation theory [8
] may play an important role to explain this mechanism. We notice that we did not see any silver nanoparticles at the bottom of SiNWs.
SEM images of sample A. (a) silver dendrites and nanoparticles and (b) magnification of silver nanoparticles.
However, in the case of sample B, where a 30-nm silver film was deposited before the etching step, we observe pores formation at a depth of 200
μm as shown in Figure . Figure a is a tilted SEM image of the sample showing both the top surface and the cross section. We notice the beginning of pores from the top of the wafer and the presence of silver nanoparticles in the bottom of pores which is magnified in image (Figure b). The formed pores are not uniform. Hence, we observe the pores at different depths. The average of the pore diameters is around several micrometers. This could be due to the thickness of the silver film.
SEM images of sample B. (a) Tilted cross section and (b) a magnification showing the silver particles.
From the obtained results for samples A and B, we estimate the etching rate for sample A at about 0.5
μm/min and the penetration rate of silver particles for sample B at about 2.7
In this work, we aim to explain the mechanism governing silicon etching to obtain SiNWs or pores. In the two metal-assisted etching structures, the intrinsic properties of the metal orientate the silicon etching. The latter is initiated by the formation of silicon dioxide. It was reported [9
] that silver-assisted etching of silicon, in presence of H2
, leads to a homogenous and large film SiNWs. In addition, the etching reaction is 250 times lower in absence of H2
]. However, as obtained in the present work and reported by Peng et al. [1
], SiNWs could be also obtained without H2
. In the used solution in this work, there are many oxidizing agents, and the most important one is the (NO3
ions. The principal reactions explaining the silicon dissolution are the following:
The electroless metal deposition mechanism was presented by Morinaga et al. with copper particles [10
]. Then Peng et al. started with the protecting mechanism that silver has during the silicon etching in the one-step method [1
]. After that, they provided in [11
] their new position regarding the etching processes of silicon wafers in aqueous HF/AgNO3
solution where the catalyzing effect was described. Qiu et al. [12
] gave experimental evidences that silver protects the silicon underneath. Consequently, the role that silver has was still not elucidated.
In this section, we give the explanation of SEM images in Figures and . Therefore, to understand the etching process of samples A and B, we propose a unique interpretation of the experimental observations in both cases by analyzing what happens at the silver/silicon interface. In Figure , we give the schematic band structure at the silver/silicon interface. In such a case, the work function of silver (ϕAg
eV) is less than that one of silicon (ϕSi
eV). Consequently, at contact, electrons diffuse from the metal to the semiconductor until the thermodynamic equilibrium is reached at the same Fermi level.
Schematic band diagram at the Ag/Si interface.
The scheme in Figure explains the Ag/Si interface in the case of evaporation (sample B). However, using work functions of Ag and Si becomes impossible when the silver is at its ionic form. This is related to the fact that the work function of a metal is defined as being the needed energy to extract one electron from its metallic conductor. Unfortunately, this is not applicable in the case of silver ions when they capture electrons from silicon. In such a case, we have to use electronegativity which is the aptitude of an element to attract electrons. In Table , we give values of the electronegativity of Ag, Si, O, and B (boron) elements involved in the etching mechanism. The electronegativity of the silver ion is greater than that of silicon. The electrons move from silicon to silver ions. At contact, the silicon underneath the silver particles becomes more positively doped than the silicon around silver.
Table 1 Values of electronegativity of Ag, Si, O, and B
The aptitude of oxidizing silicon depends on the silicon status; from values in Table , we can deduce that oxidizing pure silicon substrate (intrinsic silicon) is easier than oxidizing pure boron material. We can extrapolate that boron-doped silicon has an effective electronegativity value ranging between 1.90 (Pure Si) and 2.04 (Pure B). Consequently, in the case of sample A where silver ions are deposited on the Si substrate, there is transfer of electrons from Si to silver. Therefore, the region underneath silver becomes poor in electrons as compared to the region around silver nanoparticles. Hence, oxidizing Si around silver particles (P-doped Si) becomes more important than the Si underneath it, which behaves like P++-doped Si. Thus, the etching of oxidized sites around silver particles starts leading to the formation of SiNWs (Figure A, processes a-c). However, in the case of sample B, due to the neutral region formed underneath the evaporated silver nanoparticles, Si around them (P-doped Si) is a difficult oxidized region as compared to Si underneath them that behaves like intrinsic Si. As a result, etching starts underneath Ag particles leading to pores formation (Figure B, processes a-c).
Schematic illustration of the etching mechanism. (A) the one-step process and (B) the two-step process.
To confirm the above mechanisms, we use twin samples (sample C). After cleaning, samples were immersed for a short duration (3
min) in a HF/AgNO3
solution with the same concentrations used for sample A. In Figure , we show a top SEM image of one sample where we observe silver nanoparticles deposited in the HF/AgNO3
(image a). However, image b is a cross-section SEM view of the second sample after removing the silver film, where we observe the beginning of silicon etching. Then, the sample C with silver nanoparticles was immersed in a 40% HF aqueous solution, as done for sample B, for 2
min. Figure is a SEM image of this sample after etching, where we observe silver nanoparticles in the bottom of etched silicon.
Top and cross-section SEM images of sample C (a) chemically deposited Ag film and (b) after removing the Ag film.
Top SEM image of sample C after etching in HF solution.
According to our assumptions on the role of silver during the silicon etching, the one-step process starts during the 3
min in the HF/AgNO3
solution. The two-step process starts when the sample is immersed in sole aqueous HF solution, since silver nanoparticles are deposited in the HF/AgNO3
solution. In Figure , we give the cross-section SEM image of sample C (with silver nanoparticles) where we can see that no silver nanoparticles are found in the bottom of pores. However, the Ag nanoparticles are found on the top (white color) which confirms the one-step process. In addition, it is clear in our observation that silver nanoparticles start their catalyzing effect (the two-step mechanism) by digging silicon at depths of 1–2
Cross-section SEM image of sample C showing both protecting and catalyzing effect of Ag nanoparticles.