J. Phys. Chem. C 2008, 112, 1533-1538
1533
Etching of n-Si(111) in 40% NH4F Solution Investigated by OCP, In Situ EC-STM, and ATR-FTIR Spectroscopic Methods Sang-Eun Bae, Jung-Hyun Yoon, and Chi-Woo J. Lee* Department of AdVanced Materials Chemistry, Korea UniVersity, Jochiwon, Choongnam 339-700, Korea ReceiVed: August 19, 2007; In Final Form: NoVember 1, 2007
The etching dynamics of a moderately doped n-Si(111):H electrode misoriented in the [112h] direction in 40% NH4F aqueous solution was investigated by open circuit potential (OCP) measurement, in situ electrochemical-scanning tunneling microcopy (EC-STM), and attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy. It was demonstrated that OCP could be used as an in situ probe in judging that the well-defined Si(111):H surface was prepared successfully in the 40% NH4F solution. The OCP initially decreased when an oxide-covered silicon wafer was immersed into the solution, increased, and finally reached a plateau value, where the EC-STM image was observed to be composed of the atomically flat terraces with a threefold symmetry of interatomic distance 3.8 Å separated by a step of 3.1 Å. ATR-FTIR spectra showed that a strong and sharp peak with a full width at half-maximum of 0.94 cm-1 was observed from p-mode and a simultaneous background-like signal was observed from s-mode, suggesting that the surface chemical bond of the silicon wafer was mainly H-Si(111) normal to the surface with negligible contribution from other hydrogens bonded to surface silicon atoms. Time-dependent EC-STM images at eight different potentials including OCP revealed that the site dependence in the removal of surface silicon atoms was apparent at potentials up to the OCP and led to lateral etching with straight-step shaped terraces maintained. The first set of etching rate data was reported for the interfacial reaction at the n-Si(111):H under different potentials in the saturated fluoride solution. The data were rationalized within the framework of the recent report of the extended kinetic Monte Carlo simulation (Zhou, H.; Fu, J.; Silver, R. M. J. Phys. Chem. C 2007, 111, 3566).
1. Introduction Since Chabal discovered that basic fluoride solutions produced ideally terminated Si(111) surfaces with silicon monohydride oriented normal to the surface with low defect density,1,2 Si(111):H surfaces have been used as model surfaces for silicon research in finding an alternative cleansing solution as well as in building nanostructures for charge-transfer studies in nanoelectronics.3,4 Although the atomically flat Si(111):H surface has been used as a starting substrate of these researches, structural changes occurring during the preparation step are not well-defined yet. We have been interested in uncovering the dynamic behavior of n-Si(111):H in 40% ammonium fluoride solutions.5,6 To delineate the processes involved in the etching, scanning tunneling microscopy (STM) would be the best available technique in terms of resolution in real space. Behm discovered strongly anisotropic rates for the etching of n-Si(111) surfaces in alkaline fluoride solutions from ex situ STM observation of atomic structures in vacuum and that slow nucleation of vacancies in monohydride-terminated (111) terraces was followed by rapid etching of newly exposed dihydride Si atoms, which resulted in a “step flow” etch mechanism where extended (111) terraces were eroded along the most stable [11h0] steps with {111} microfacets.7 To directly probe the processes involved in the etching at the interface of silicon/electrolyte solution, in situ electrochemical scanning tunneling microscopy (EC-STM) would be required * Author to whom correspondence should be addressed. E-mail: cwlee@ korea.ac.kr; phone: +82-41-860-1333/+82-2-3290-3990; fax: +82-41-8676823/+82-2-928-1330.
to be employed to keep the electrode potential at a constant value during scanning. Itaya reported from in situ EC-STM study in dilute NH4F solution that multiple hydrogen-terminated Si atoms at the kink and step sites were eroded more rapidly than the monohydride Si step, that the density of kinks played a main role in controlling the etching rate of Si, and that the difference in the reactivity guided the dissolution of Si in a layerby-layer fashion.8 Although 40% NH4F solution has been the choice of medium in obtaining ultraclean atomically flat Si(111):H surfaces, Ye was the first by employing in situ ECSTM to investigate the dynamics of etching process on an n-Si(111):H in 40% NH4F solution and measured the etching rate as 86 and 28 nm/min for dihydride and monohydride Si(111) steps, respectively, at the potential near open circuit potential (OCP).9 Recently Silver simulated Ye’s work to generate the seven parameters including defect etch rate and defect concentration10 by extending the kinetic Monte Carlo (KMC) method, which has been used successfully by Hines to simulate steady-state STM images.11,12 To better understand the etching dynamics of the model substrate in 40% ammonium fluoride solution, more quantitative information on the etching rates should be provided by using in situ EC-STM in the medium. At the present time, Ye’s work is the only measurement of the kind, which concerned the Si(111) misoriented in the [1h1h2] direction and dealt with sawtooth shaped terraces, instead of more common straight-step shaped terraces, at the two different potentials including the potential near OCP.9 In this work, we investigated the etching of the Si(111) misoriented in the [112h] direction in the saturated aqueous solution of NH4F by using in situ EC-STM at eight
10.1021/jp076673m CCC: $40.75 © 2008 American Chemical Society Published on Web 01/10/2008
1534 J. Phys. Chem. C, Vol. 112, No. 5, 2008
Bae et al.
different potentials together with OCP and ATR-FTIR measurements and wish to report the first set of data directly measured by in situ EC-STM for the etching rates of n-Si(111):H of straight stepped terraces in the fluoride solution. The result was interpreted by using Silver’s extended KMC work.10 2. Experimental Section Single-side polished silicon (111) wafers (n-type, 1-12 Ωcm, 18 MΩcm) and acetone, oxidized in a H2O2:H2SO4 (1:1) solution for 20 min, rinsed with a Milli-Q water, and finally dried in the flowing purified nitrogen gas. The oxide covered silicon wafer was immersed into or contacted with 40% NH4F solutions for OCP measurements. The EC-STM experiment was carried out by using a PicoSPM and PicoStat (Molecular Imaging Corp.). A Teflon EC-STM cell and a chamber were designed and made for the proper operation of EC-STM in ambient conditions, as well as for the prevention of oxygen dissolving into the solutions during all of the experiments. Ohmic contact was made to the Si by GaIn alloy. Pt wire and loops were used as the reference electrode and the counter electrode, respectively. The Pt quasi-reference electrode was calibrated with respect to a saturated calomel electrode (SCE) at the end of experiment, and all potentials are quoted versus SCE for convenience. EC-STM tips were prepared by mechanical cutting of Pt/Ir wire (0.25 mm in diameter) or electrochemical etching of a tungsten wire (0.25 mm in diameter) in 1 M NaOH solutions first at 10 V AC and then 1 V DC, and followed by insulating with Apiezon wax to minimize the Faradaic current at the tip as was used previously.13 All reactions and EC-STM scanning of silicon surfaces were conducted in oxygen-free environments at room temperature in the dark. ATR-FTIR measurement was performed with double-side polished n-Si(111) by using a Bio-Rad Excaliber spectrometer equipped with HgCdTe (MCT) detector cooled with liquid nitrogen as before.14 The pH of the 40% NH4F solution was measured by using an antimony electrode (Metrohm Co.). All of the chemicals used in this work were of a semiconductor grade, available commercially, and were used without further purification. Solutions were made of Milli-Q water. 3. Results and Discussion Although many reports have described the preparation of atomically flat Si(111):H from 40% NH4F solution, ideally hydrogen-terminated silicon surfaces were often not prepared as was thought because the surface chemistry of silicon was so elusive.15 Therefore, we wanted to follow the process during the preparation of Si(111):H in the fluoride solution by some in situ experimental probe, for which we chose OCP as we demonstrate below. Figure 1 shows changes in OCP as a function of time when the silicon (111) wafer covered with silicon oxide as prepared in the Experimental Section was contacted with 40% NH4F solution at pH 8.3. The OCP decreased sharply toward a negative potential (-1.12 V, 70 s) as soon as the silicon wafer contacted the fluoride solution and increased to a plateau value (-0.35 V, 800 s). This was due to the local equilibrium established near the electrode surface among the redox species involved because control experiments by bubbling argon through the bulk solution pushed the OCP to -0.3 V and those by bubbling oxygen returned the OCP to -1.1 V as was observed by Allongue.16 In spite of the problem whether the OCP reflected the true equilibrium potential or not, the fact that the OCP reached a plateau potential region in this
Figure 1. Change in OCP of n-Si(111)/silicon oxide electrode contacted with 40% NH4F solution under inert atmosphere and dark.
protocol could be used as important information in judging whether an atomically flat Si(111) surface was prepared from the chemical treatment of the silicon(111)/silicon oxide wafer in the fluorinated solution and we will call the plateau potential as “equilibrium” OCP here because the plateau value remained more than several hundred seconds as can be seen in Figure 1 and because the cyclic voltammogram showed zero current at -0.35 ((0.01) V. The OCP has been shown to be a sensitive experimental variable of reflecting the difference in the orientation of silicon in 10 M NH4F solution at pH 8.17 The OCP can be deduced as being identified as a redox mixed potential of cathodic hydrogen evolution and anodic oxidation of Si resulting finally in SiF4 or SiF62-, once the surface oxide on silicon wafer is completely dissolved into the fluoride solution and the interface of Si(111)/ammonium fluoride solution is established.18 Thus, it was regarded as a good in situ probe to judge whether silicon oxide on silicon was completely removed or not, namely, whether the surface of silicon wafer was free of silicon oxide to transform to ultraclean atomically flat Si(111) surfaces or not. We have checked this by means of in situ EC-STM when the OCP reached the plateau value. The EC-STM images shown in Figure 2 were obtained at the substrate potential of -0.60 V, when 800 s (marked as red in Figure 1) has passed from the time when the silicon/silicon oxide wafer contacted the fluoride solution. The potential value significantly more negative than the equilibrium OCP was applied to the silicon electrode to keep the morphology of the surface little changed (vide infra). Figure 2a shows a representative in situ EC-STM image, where a sequence of relatively equidistant, parallel, and straight steps can be seen with a regular step height. The height of 3.1 Å corresponds to steps between subsequent silicon bilayers, and the average terrace width of 40 nm on the (111) surface results from the miscut
