NANO LETTERS
A High-Density Array of Size-Controlled Silicon Nanodots in a Silicon Oxide Nanowire by Electron-Stimulated Oxygen Expulsion
2009 Vol. 9, No. 5 1780-1786
Gyeong-Su Park,*,†,‡ Eun Kyung Lee,†,§ Jun Ho Lee,‡ Juyeon Park,| Seong Keun Kim,| Xiang Shu Li,‡ Ju Cheol Park,‡ Jae Gwan Chung,‡ Woo Sung Jeon,‡ Sung Heo,‡ Jae Hak Lee,‡ Byoung Lyong Choi,§ and Jong Min Kim§ Analytical Engineering Center and Frontier Research Laboratory, Samsung AdVanced Institute of Technology, PO Box 111, Suwon, Korea 440-600, and Department of Chemistry, Seoul National UniVersity, Seoul, Korea 151-747 Received December 3, 2008; Revised Manuscript Received February 16, 2009
ABSTRACT Methods of producing Si nanodots embedded in films of silicon oxide and silicon nitride abound, but fabrication of Si nanodots in a nanowire of these materials is very rare despite the fact that nanowire architecture enhances the charge collection and transport efficiencies for solar cells and field-effect transistors. We report a novel fabrication method for a high-density array of size-controlled sillicon nanodots from a silicon oxide nanowire using electron-beam irradiation. Our results demonstrate that a highly dense phase of Si nanodots with a narrow size distribution can be made from a silicon oxide nanowire with a core-shell structure of crystalline silicon-rich oxide (c-SRO)/amorphous silicon oxide (a-SiO2). This new nanomaterial shows the carrier transport characteristics of a semiconductor. The initially produced amorphous Si nanodots can be readily turned into crystalline Si (c-Si) nanodots by thermal annealing. Key characteristics of c-Si nanodots such as their size, number density, and rate of nucleation and growth are easily controlled by varying the electron radiation dose and annealing temperature. Nanodot formation is mechanistically initiated by electron trapping at the c-SRO core as well as at the core-shell interface, which leads to out-diffusion of the negatively charged oxygen through Coulomb repulsion, fostering the aggregation of Si atoms.
The current semiconductor industry could be readily transformed into an immensely useful future technology if one could produce, on an industrial scale, Si nanodots of controllable size embedded in a SiO2 matrix.1,2 Such a system holds great promise for use in light-emitting diodes, image sensors, photovoltaic solar cells, and charge trap devices due to its unique properties of variable quantum size effects.3,4 Although an array of various sized semiconductor quantum dots such as CdSe5 has been assembled on a TiO2 nanotube, with the prospect of efficiently harvesting the spectral range to the entire visible region of solar radiation,6 no such case has been reported to date in a nontoxic, inexpensive, and easily processed silicon-based material. Fabrication of Si nanodots in nanowires is very rare despite the fact that * To whom correspondence should be addressed, gs8144.park@samsung. com. † These authors contributed equally to this work. ‡ Analytical Engineering Center, Samsung Advanced Institute of Technology. § Frontier Research Laboratory, Samsung Advanced Institute of Technology. | Department of Chemistry, Seoul National University. 10.1021/nl803660m CCC: $40.75 Published on Web 03/27/2009
2009 American Chemical Society
nanowires have been shown to improve charge collection efficiency7 and enable low-temperature processing for use in polymer blend8 and dye-sensitized9 solar cells. Here, we report a novel fabrication method for a highdensity array of size-controlled Si nanodots from silicon oxide nanowires using electron-beam irradiation. We also elucidated an unusual formation mechanism for these nanodots that involves facile electron trapping at the defective crystalline silicon-rich oxide (c-SRO) core as well as at the core-shell interface. This new mechanism, perhaps to be termed electron-trapping-induced synthesis, can be of immense value in the chemical transformation of nanomaterials and has a potential to become a useful and general paradigm. We synthesized the core-shell silicon oxide nanowires10 by simply heating in a microchamber an n-type (100) silicon wafer onto which a thin (∼20 nm) gold layer had been deposited to 1273 K. The diameters and lengths of the nanowires measured by scanning electron microscopy (SEM) images were in the range of 40-120 nm and a few tens of micrometers, respectively. To understand how our core-shell nanowires came to be formed, we studied their structural
Figure 1. Structure and composition of the core-shell silicon oxide nanowire. (a) TEM image of a core-shell silicon oxide nanowire with a core diameter of 12 nm. The areas marked A1 and A2 indicate the core region and shell region of the nanowire, respectively. The inset shows a low-magnification TEM image of the typical core-shell nanowire. (b) HR-TEM image of the square-framed area shown in (a). (c) Selected area electron diffraction pattern of the same area as (b). (d) Core-excitation EELS spectra taken at areas A1 and A2 as well as from the reference Si wafer. The sharp onset at 99.8 eV (indicated by an arrow) corresponds to the Si-L2,3 transition. (e) EDS spectra measured at areas A1 and A2, which show different stoichiometries (SiO1.1 for A1 and SiO2.0 for A2). (f) Fourier transform infrared spectrum of nanowire samples. Two absorption peaks at 800 and 1080 cm-1 correspond to the bending and stretching modes of Si-O-Si vibration, respectively.
evolution through detailed transmission electron microscopy (TEM) and SEM analyses, which revealed that the intrinsic core-shell structure of the silicon oxide nanowires was generated by the segregation of silicon from the supersaturated gold silicide droplets (Supporting Information, Figure S1). A typical TEM image of the core-shell silicon oxide nanowire is shown in Figure 1a. The diameters of the core are in the range of 5-15 nm depending on the growth conditions. Diffraction contrast of the core region is consistent with a crystalline structure, whereas the bright contrast observed in the shell region is indicative of an amorphous structure. The TEM image of the core-shell structure shows a smooth interface between the crystalline core (c-core) and the amorphous shell (a-shell). The lattice-resolved TEM image (Figure 1b) and electron diffraction pattern (Figure 1c) obtained at the square-framed core region of Figure 1a further show that the core consists mostly of a single-crystal structure with some oxygen vacancies and stacking faults. Atomic distances measured by electron diffraction indicate that the c-core structure is close to the cubic-type SiO2, although there is some atomic displacement due to the defects. Figure 1d shows the core-excitation electron energy loss spectra (EELS) taken at the c-core (A1) and a-shell (A2) along with EELS for pure Si. With the Si-L2,3 energy loss near edge structure (ELNES) known to occur at the onset of 99.8 and 106 eV for Si and SiO2,11 respectively, it is clear that the composition of the a-shell region (A2) is mainly Nano Lett., Vol. 9, No. 5, 2009
that of SiO2 whereas the c-core (A1) appears to be a mixture of Si and SiO2,12 which is to be called silicon-rich oxide (SRO) hereafter. In Figure 1e, energy dispersive X-ray spectroscopy (EDS) reveals a stoichiometry of SiO1.1 for A1 and SiO2.0 for A2, in full accord with the EELS prediction. Fourier transform infrared spectroscopy shows that the nanowire samples used in this experiment have only the Si-O-Si bond with no other types of bonds such as Si-O-C or Si-OH (Figure 1f).13,14 A crystalline SRO phase is unusual, and a c-SRO/a-SiO2 core-shell structure is even more unexpected. Irradiation of the core-shell nanowire by an electron beam produced a high density of nanodots. Energy-filtered transmission electron microscopy (EF-TEM) gave the plasmon loss images of Si filtered at 17 eV (Figure 2a,b, for two different diameters of nanowire),15,16 which show that the number of Si nanodots formed increases with irradiation time. The second images of each figure show that Si atoms (red arrows) are initially aggregated around the c-SRO core and sheathe the c-SRO core with a thickness of about 1-1.3 nm (shown by the green ring in the second inset of Figure 2a). As the irradiation time becomes longer, more Si atoms aggregate into nanodots in the a-SiO2 matrix until eventually a high density of Si nanodots that are 2-4 nm in size are formed in the a-SiO2 shell. The EF-TEM images of the core-shell nanowire obtained before and after electron-beam irradiation indicate that the 1781
Figure 2. Formation of Si nanodots in the c-SRO/a-SiO2 core-shell nanowires during the electron-beam irradiation at room temperature. The electron energy was 300 keV and the beam intensity was 1.2 A/cm2. Irradiation-time-dependent plasmon loss images (filtered at 17 eV for Si) of a 76 nm diameter core-shell nanowire with a c-SRO core diameter of 10.5 nm (a) and a 120 nm diameter nanowire with a core diameter of 9.5 nm (b). The insets in the first images show the TEM image of each nanowire before irradiation. The circular diagrams schematically depict a cross-sectional view of the core-shell nanowire during irradiation, which represents the intrinsic c-SRO core (red), the Si-sheath around the core (green), the Si nanodots (black), and the a-SiO2 shell (pink). Due to the larger diameter and thus the sample thickness, the EF-TEM images of (b) suffer from a lower contrast. (c) Si plasmon loss image of a 49 nm diameter core-shell nanowire with a c-SRO core diameter of 6.5 nm after electron irradiation for 85 min at the same electron beam intensity. The image shows a trace of Si-rich phase in the nanowire. The inset shows the plasmon loss image filtered at 21 eV for the c-SRO core. (d) HR-TEM image of the 76 nm diameter nanowire taken after 70 min of irradiation. (e) Selected area electron diffraction pattern of the same sample as (d).
c-SRO core remains (Supporting Information, Figure S2) but the a-SiO2 shell undergoes a significant volume reduction, which may have resulted from the “evaporation” of the more volatile species of the a-SiO2 matrix, i.e., the O atom, following the dissociation of the Si-O bond. Although the mechanism for such dissociation and out-diffusion of the O atoms is to be discussed later, their expulsion would certainly result in volume contraction of the nanowire as well as enrichment of the Si atoms, which would lead to the aggregation of Si into clusters and nanodots. On the other hand, if the nanodot formation had been caused by the electron-beam heating,17,18 we would expect that a narrower nanowire could be more favorable in the nanodot formation. However, a close examination of the EF-TEM images for nanowires of various overall and core diameters (Figure 2a-c) found that the nanodot formation was dependent not on the overall diameter but on the core diameter. For example, even a very thick nanowire of 120 nm diameter is shown to form a high density of Si nanodots when it has a sufficiently large core diameter of 9.5 nm (Figure 2b). 1782
We found that the diameter of the c-SRO core determines the nucleation rate and growth pattern of Si nanodots regardless of the overall nanowire diameter, which indicates that the c-SRO core plays a vital role in the formation of Si nanodots. The high-resolution TEM (HR-TEM) image and electron diffraction pattern of the 76 nm diameter nanowire taken after irradiation show that the Si nanodots are amorphous and the c-SRO core remains unchanged (Figure 2d,e). We found that a pure a-SiO2 nanowire (with no c-SRO core) failed to produce Si nanodots under the exact same experimental conditions. Instead, only damage induced by the electron beam was observed, which showed a large extent of deformation of the nanowire structure with time (Supporting Information, Figure S3). Although there have been studies that reported the transformation of a-SiO2 film into a-Si following Si-O bond breakage and Si-Si bond formation through mechanisms such as knock-on displacement,19 electron-stimulated desorption,20 and valence electron excitation,21 they do not fully explain the role of the c-SRO core in the transformation of Nano Lett., Vol. 9, No. 5, 2009
Figure 3. Effect of accumulated electric charges on the growth of Si nanodots and electrical conduction through the core-shell nanowire. (a) Si plasmon loss image of a 50 nm diameter nanowire with no supporting film underneath after 35 min of electron-beam irradiation at 4.8 A/cm2. (b) The same image as (a), except that the nanowire was placed on a conductive carbon film during irradiation, which drained the charge accumulated in the nanowire. (c) Si plasmon loss image of a 67 nm diameter nanowire after 35 min of electron-beam irradiation at 8.2 A/cm2 at room temperature. (d) The same image as (c), except that the irradiation was done at 573 K. For (a-d), TEM images of each original core-shell nanowire are shown in the insets. (e) TEM image of a 61 nm diameter core-shell nanowire in contact with an STM tip (Pt-Ir) inside TEM. (f) Si plasmon loss image of the nanowire in (e) after 5 min of electron-beam irradiation at 25 A/cm2. The nanowires before and after electron-beam irradiation are labeled A and B, respectively. (g) Current vs voltage (I-V) curve for nanowire A (with no Si nanodots formed within) over the voltage range of -70 to +70 V. (h) I-V curves for the two nanowires A and B with a bias voltage of up to 25 V. For (g) and (h), the contact between the Pt-Ir probe and the nanowire was constantly monitored inside TEM.
a-SiO2 shell into a-Si at room temperature under electronbeam irradiation. We therefore carried out further experiments to elucidate the role of the c-SRO core in the formation of Si nanodots, with particular attention paid to the effect of the enormous amount of electron charges supplied by the electron beam. First, a set of two 50 nm diameter nanowires with nearly the same c-SRO core (7.6 nm in diameter) were irradiated for 35 min using an electron-beam intensity of 4.8 A/cm2. One nanowire was irradiated without a supporting film underneath (Figure 3a) while the other was placed on a conductive carbon film (Figure 3b) during irradiation. Comparison of the two EF-TEM images shows a striking Nano Lett., Vol. 9, No. 5, 2009
difference in the number density of Si nanodots formed, with only a trace of the Si-rich phase in the latter case (indicated by an arrow in Figure 3b) in the vicinity of the c-SRO core. This result strongly suggests that the drainage of accumulated electric charge in the silicon oxide nanowire through the conductive carbon film hinders the transformation of a-SiO2 into a-Si. It is well-known that thermalized electrons are generated when a dielectric is irradiated with high-energy electrons and that they are trapped in disordered structures,22 which may include the core-shell interface and the defective c-SRO core of our nanowire sample. A large number of trapped electrons will affect the chemical environment of 1783
Figure 4. Formation of c-Si nanodot by thermal annealing. (a) Si plasmon loss image of a 60 nm diameter core-shell nanowire after 25 min of electron-beam irradiation at 10.4 A/cm2. The inset shows the zero-loss image of the core-shell nanowire. (b) The same image of nanowire (a), taken after thermal annealing at 893 K for 45 min. The inset shows the zero-loss image of the core-shell nanowire. (c) HR-TEM image of nanowire (a), showing no lattice fringes. (d) HR-TEM image of nanowire (b), which clearly shows the formation of crystalline domains of Si nanodots. (e) HR-TEM image of the nanowire a after 30 min of thermal annealing at 823 K. (f) HR-TEM image of the nanowire a after 15 min of thermal annealing at 973 K. Larger domains were produced at a higher temperature despite the shorter time for annealing. (g) Low-energy loss spectra (with a spatial resolution of 1 nm) taken comparatively at the three regions, namely, the a-SiO2 shell, the a-Si nanodot, and the c-Si nanodot.
the nanowire since the negatively charged oxygen atoms are going to be placed in the electric field of these trapped charges and thus experience a repulsive interaction, which will eventually facilitate the out-diffusion of oxygen as the Si-O bonds break. Since the enrichment of Si following the expulsion of oxygen appears to start from the vicinity of the c-SRO core at a rate proportional to the core diameter, the electrons generated by the electron-beam irradiation are expected to locally accumulate at the c-SRO/a-SiO2 interface and at the defective c-SRO core. Second, a 67 nm diameter nanowire was irradiated at room temperature (Figure 3c) and at 573 K (Figure 3d) for 35 min under the same electron beam intensity. The EF-TEM image shows virtually no nanodots in the sample irradiated at 573 K, in stark contrast to the usual case at room temperature, which is likely to be due to the thermionic emission of the trapped electrons23 at high temperature. The results shown in Figure 3a-d are strong and direct evidence that the formation of Si nanodots 1784
in the core-shell nanowire is promoted by electron trapping at the c-SRO/a-SiO2 interface and at the c-SRO core during the electron-beam irradiation. Electron transport properties of the c-SRO/a-SiO2 core-shell nanowire before and after Si nanodot formation were investigated using a scanning tunneling microscope (STM) installed in a TEM holder. The TEM image of Figure 3e shows that the c-SRO core was 7 nm in diameter and the distance between the probe and the electrode was 215 nm. After contact was made between the Pt-Ir probe and the nanowire, we generated Si nanodots in the latter by using electron-beam irradiation at 25 A/cm2 for 5 min. Figure 3f gives the plasmon loss image for Si recorded after the Si nanodot formation, which clearly shows that a high number density of Si nanodots was successfully formed in the nanowire. The resulting volume contraction was estimated to be 9.7% of the initial a-SiO2 shell volume. When we apply a maximum bias voltage of (70 V to the original nanowire Nano Lett., Vol. 9, No. 5, 2009
Figure 5. A high-density array of size-controlled Si nanocrystals in a silicon oxide core-shell nanowire. (a) Magnified TEM images of c-Si nanodots obtained at three framed areas of a nanowire surface. (b) Si plasmon loss image of the silicon oxide nanowire taken after arranging the different sized Si nanocrystals in an orderly pattern using the thermal annealing conditions of Figure 4. (c) Schematic representation of the new architecture for different sized Si nanocrystals.
labeled A (Figure 3e), we obtained the current-voltage (I-V) curve shown in Figure 3g. It has a broad bump24 in current that suggests electron accumulation in the core-shell nanowire due to high voltage stress. This result is another demonstration that charge trapping is possible in the original core-shell nanowire. Figure 3h shows the difference in the I-V curve of the nanowire before and after the Si nanodot formation. Despite the contact resistance between the probe and the nanowire, a maximum current of 15 nA measured at 24 V across the nanowire (i.e., with a resistance of ∼1.6 GΩ) that contains the Si nanodots is remarkable in comparison with the original nanowire that exhibits the characteristics of an insulator. The dramatic electrical transition from an insulator to a semiconductor may be due to the transport of electrons through the Si-rich phase surrounding the c-SRO core as well as the sequential electron tunneling through the a-Si nanodots. It is also to be noted that conduction displays a staircase-like behavior25,26 in the I-V curve with a sudden change in current. This may be because each Si nanodot isolated from the others in the a-SiO2 matrix acts like a Coulomb island with a tunnel junction and thus exhibits the Coulomb blockade effect.26 In order to determine the number density of a-Si nanodots in a typical electron-beam-treated core-shell nanowire, we reconstructed a three-dimensional image for the distribution of the Si nanodots by combining plasmon-filtered microscopy with electron tomography (movie in Supporting Information).16,27 The EF-TEM tomography image reveals several key features, namely, (1) a dense layer of Si forms a sheath of 1.2 nm thickness around the c-SRO core; (2) larger nanodots (2-4 nm) are mainly found near the nanowire surface; and (3) the number density of Si nanodots in the Nano Lett., Vol. 9, No. 5, 2009
a-SiO2 shell is at least 5 × 1018 cm-3. Small Si nanodots less than 2 nm in size were excluded from the calculation due to their low intensity in the tomography image. If we assume the same nanodot number density for the nanowire labeled B in Figure 3f, the number of Si nanodots in this nanowire should be about 3500. On the other hand, if we assume that the EF-TEM-measured volume contraction of 9.7% for the nanowire B is solely due to the out-diffusion of oxygen, we can also estimate the number of Si nanodots assuming an average nanodot size of 3 nm, which is about 8000. These two values are in reasonable agreement given the approximate nature of the estimation and the discard of smaller nanodots. If each a-Si nanodot can trap as many as four to five electrons as is generally known,28 the maximum number of electrons trapped in Si nanodots in the nanowire labeled B will be in the order of 104-105. Figure 4 compares the Si plasmon loss images and HRTEM images of a core-shell nanowire (electron-beam irradiated at 10.4 A/cm2) before and after thermal annealing. The as-fabricated Si nanodots (Figure 4a,c) are 2-4 nm in size and amorphous, but upon annealing at 893 K for 45 min, they become larger in size (Figure 4b) and crystalline with a typical domain size of 4-6 nm (Figure 4d); at the same time small nanodots (
