The metal-assisted chemical etching of silicon on p-types of silicon wafer was performed in AgNO3/HF solution. The growth mechanism of silicon nanowires and silver dendrites was investigated by scanning electron microscopy. The formation and growth mechanism of the silver dendrites is explained based on the diffusion-limited aggregation model. Weight lost measurements and SEM images showed that the etch rate of silicon increased slightly with time. The etching mechanism was explained based on the oxidation of silicon at the silver-silicon interface and the transfer of electrons between the silicon wafer and silver dendrites.
Keyword: Silicon nanowires, metal-assisted chemical etching, silver dendrites
Silicon is the most important semiconductor material in the electronics industry and silicon nanowires (SNWs) have attracted much attention, because of their potential applications in nanoscale optoelectronics [1, 2] and electrochemical devices [3, 4]. Various methods of preparing one-dimensional silicon nanostructures have been developed [5-8]. However, these growth mechanisms have some limitations as they generally need a high temperature or a high vacuum, templates and complex equipment, or they employ hazardous silicon precursors.
Recently, the metal-assisted chemical etching (MACE) technique, which is used to produce SNWs, has attracted a great deal of attention from many researchers [9-16], because of its simplicity and low cost. This technique has been widely used to manufacture silicon wire for many different applications, such as thermoelectric materials , photoelectrochemical solar cells , and lithium-ion battery anodes , the mechanism of the etching process still remains controversial . Among the metals used as the catalyst for the etching process, silver is the best candidate, because it is inexpensive and can produce SNWs with high uniformity and highly efficient etching. The metal-assisted chemical etching reactions, in general, can be classiï¬ed into two types, one-step reaction in etchant solution containing HF and metal salts  and two-step reaction that involves the predeposition of metal nanoparticles [19, 20] followed by chemical etching in the presence of HF and H2O2. In one-step reaction method. Although it was affirmed that the growth of silver dendrites plays an important role in the etching process [10, 13, 21, 22], further research on the mechanism of electronic transport and the growth processes of the silver dendrites are still needed. In this paper, the MACE method was used to etch p-types of silicon wafer (100) in order to fabricate SNW arrays. Based on the experimental results, The mechanism of chemical etching of silicon assisted by silver particles has been studied.
2. Experimental procedure
Si (100) wafers (p-type, Boron doped; 10-30 Ωcm; 500 µm thick) were used in this work. HF (49%), H2O2 (30%), HNO3 (aq. 60-61%), H2SO4 (98%), AgNO3 (>99.9%) were purchased from Sigma-Aldrich.
The Si wafers were cut into 2x2 cm2 pieces and ultrasonically degreased in acetone and ethanol successively at room temperature for each 20 minutes. The wafer pieces were then cleaned in a boiling Piranha solution (H2SO4:H2O2 = 4:1; vol/vol) for 1 hour, rinsed thoroughly with deionized water for 10 min and then dipped into 1% HF solution for 1 min. After that, the cleaned silicon wafers were separately etched in two solutions at room temperature for different periods of time. One solution contained 0.02 mol/l AgNO3 and 4.6 mol/l HF (solution 1) and the other solution contained 0.01 mol/l AgNO3 and 4.6 mol/l HF (solution 2). The container employed was a conventional Teï¬‚on-coated stainless steel vessel with a volume of 2 l. After the etching process, the silicon wafers were rinsed with deionized water and blown dry by air. The silver dendrites was removed from the etched silicon wafer by soaking it in a 25% HNO3 solution at room temperature for 1 hour. To perform the weight measurements, the samples were dried in nitrogen and and then weighed by a model OHAU-PA114 Pioneer Analytical Balance. The morphology of the samples was observed using a NOVA NanoSEM200 ï¬eld emission scanning electron microscope. All of the measurements were performed at room temperature.
3. Results and discussion
Figure 1 shows the cross-sectional SEM images of the Si wafers etched in solution 1 respectively for 30, 60 and 90 min, and then immersed in 25% HNO3 solution for 1 hour. It can be seen that the well-aligned vertical Si nanowire arrays were readily obtained. The diameters of the SiNWs can be observed to be in the range of 40-300 nm.
Figure 2(a) shows the top view of the Si wafer in figure 1(b). In this figure, the SNWs are clustered together at the tip and the distribution of SNWs was not clear. For further observation, the Si wafer after etching was subjected to ultrasonic treatment for 10 minutes. Figure 2(b) shows the top view of Si wafer after treatment by ultrasonic wave. Almost SNWs was cracked. The diameters of the SiNWs observed in figure 2(b) agreed with those obtained from figure 1(b). However, some silicon wires with diameters of approximately 1 µm were also observed. These wires were attributed to the agglomeration of SNWs with a smaller size and length than the others. Figure 2(b) also showed the separation of the SiNWs from each other.
Figure 3 shows the etch rate as a function of the etching time in Si wafer. Data in the plot was calculated from SEM images using image analyzer. In parallel with SEM observation, the weight loss also was used to determine the etch rate. The weight loss present in figure 3 was normalized to 1 gram of silicon. From the plots, it can be clearly seen that the etching length increased quite linearly with etching time in the period from 15 to 60 min. This results seem to be similar with the experiment that performed before by K. Peng and coworker . However, when etching time increased to 90 min, etch rate increased with etching time.
The charge transfer mechanism between a semiconductor and electrolyte solution can be explained by the Gerischer model in the case of the interface between a semiconductor and electrolyte solution . According to this model, silver particles will be deposited onto the silicon wafer based on the galvanic displacement of silver by silicon, in which two simultaneous processes occur at the silicon surface: the cathodic reaction (reduction of Ag+ ions - equation 1, which produces metallic Ag deposits) and the anodic reaction (oxidation of silicon - equation 2, silicon dissolves into solution)
(1) E0 = 0.7991 V
(2) E0 = 1.37 V
On the fact, the dissolution process of silicon in HF solution is quite complex and much research has been conducted until now . However, equation (2) can be used as the overall equation for the dissolution process.
Immediately after the deposition process, the silver particles will catalyze the etching process of the silicon wafer in AgNO3/HF solution. To investigate the additional effects of the silver particles on the etching process, as well as on the growth of the silver dendrites, we decreased the concentration of AgNO3 to slow down the reaction rate. Figure 4(a) shows the SEM image of the top view of the Si wafer etched in solution 2 for 1 second. It can be seen that silver particles with diameters of 30-90 nm were immediately deposited on the Si surface. As time goes by, as showed in figure 4(b), the silver particles become bigger and connected to each other. Figure 4(c) shows the cross-sectional view of the Si wafer etched in solution 2 for 60 min. It can be seen that some of the silver dendrites were grown from the bottom holes that originate from the etching process. These silver dendrites are connected to each other and cover the top of the silicon nanowire array.
Although the etching mechanism remains unclear, we agree with the mechanism based on the catalytic activity of the silver particles on the silicon surface that was proposed in . It is well known that the catalytic oxidative power of silver results from the fact that atomic oxygen fits into its octahedral holes . Once in contact with the silicon surface, the silver particles catalyze the oxidation process of silicon by capturing electrons from it, resulting in the dissolution of the silicon underneath the silver particles in HF acid.
As can be seen from figure 4(c), the silver dendrites grow towards the solution. This observation seems to be in contradiction with the theory proposed by T. Qiu et al . In their study, the authors believed that the silver dendrites were built up from nanoclusters produced at the bottom holes which then diffuse along the pore channels due to the concentration gradient. Follow this explanation, it is can not explain the increase of etch rate by time. Moreover, in papers published more recently [27-28], based on a Monte Carlo simulation associated with experiments, J. Fang et al. concluded that the growth of the silver dendrites is strongly dependent on the concentration gradient. According to these authors, the big silver dendrites always grow quickly and stay continuously in contact with the region of high silver ion concentration, while around the small dendrites, the growth rate is limited because of the insufficient supply of silver ions. Based on these arguments, it can be concluded that the growth of the silver dendrites results from the deposition of Ag+ ions from solution.
From the arguments presented above, we believe that the etching process and the growth of the silver dendrites take place as follows: Initially, silver particles are deposited on the silicon wafer based on the process of electroless metal plating. In the case of a silicon wafer immersed in a solution of AgNO3, the silver particles will become larger and connect to each other, resulting in a network of silver on the silicon surface. At the same time, the etching process catalyzed by silver particles takes place, resulting in the sinking of the silver-network further into the silicon wafer. This etching process of the silver-network into silicon is known to be possible, as it was observed by H Fang et al. . In the nucleation process, some of the silver particles will grow into the silver dendrites following the diffusion-limited aggregation model . Since a concentration gradient exists between the silicon surface and solution, these silver dendrites will grow faster and consume almost all of the electrons that the silver-network obtains from the etching process. The consumption of electrons by the silver dendrites and the existence of a concentration gradient will restrict the nucleation at any other position at silver-network. Finally, the silver-network continuously sinks further into the silicon wafer and the silver dendrites grow larger and larger. It should also be noted that during their growth process, the silver dendrites can become connected to each other, resulting in the formation of a network of silver dendrites. From this time, it seems that the network of silver dendrites receives the electrons supplied by the silver network embedded in the Si nanowire array. Because in a conducting network, every position is equal from an electrical point of view, the etching rate is uniform over the whole surface of the silicon wafer. Based on the above explanation, it also can be seen that the expansion of the silver dendrites will promote the deposition of silver ions on the silver dendrites. This process demands more electrons from the silicon wafer, resulting in the increase of the etch rate.
SiNWs were prepared via a metal-assisted electroless chemical etching method. From experimental results, it was concluded that the etching process is closely related to the catalytic activity of the silver particles. During the etching process, the concentration of silver ions has a strong effect on the etching rate. We also believe that a network of silver particles connected together exists on the silicon surface. This network receives electrons from silicon, transfers them to the silver dendrites and, finally, the electrons transferred from the silver dendrites to nearby silver ions result in the continuous growth of the silver dendrites. Although a growth mechanism of the silicon nanowires and silver dendrites was postulated in this paper, more extended work including experiments and theoretical calculations is needed to clarify the real mechanism of the etching process.