The nucleation stage in electron beam induced deposition

The nucleation stage in electron beam induced deposition

1

Delft University of Technology, Fac. Applied Physics, Lorentzweg 1, 2628 CJ Delft, The Netherlands

2

Center for Solid State Science, Arizona State University, Tempe, AZ 85287-1704 E-mail: C.W.Hagen@TUDelft.nl

Abstract. With electron beam induced deposition structures as small as 1 nm can be deposited.

The nucleation stage of the growth of such deposits is studied in dependence of the substrate material and substrate temperature. On amorphous carbon foils larger deposit sizes are obtained at higher substrate temperatures during growth. On SiN membranes unusually high growth rates are observed which seem to relate to charging effects.

1. Introduction

Electron beam induced deposition (EBID) is recently drawing attention as a potential nanofabrication technique. In an increasing number of groups it was demonstrated that nanostructures even down to 1 nm in size can be fabricated using EBID [1-5]. The principle of EBID is that precursor molecules, adsorbed on a substrate surface, are dissociated as a result of the interaction between an incident electron beam and the gas-substrate system. The ideal precursor consists of gas molecules that contain the material to be deposited, a metal for instance, and some ligands which are easily separated from the metal atom by the interaction with electrons, to form volatile fragments that are easily pumped away. As electron beams can be focused to 0.1 nm in today’s state of the art electron microscopes, ultra-high resolution patterning becomes possible. The resolution is for a large part determined by the size of the secondary electron emission area around the point of impact of the primary electron beam. Using Monte Carlo simulations of the electron-substrate interactions and the electron induced precursor dissociation it was predicted that this area is so small that a spatial deposition resolution better than 1 nm can be achieved [6]. However, other factors as the interaction between the precursor and the substrate including adsorption and desorption and diffusion, were not taken into account, but may play a role in the spatial resolution and the growth rate as well. When aiming at sub-nm deposits, i.e. the nucleation stage of the growth, the substrate surface topography will become important, as well as the surface condition. In this paper experimental results are presented of the first stage of the growth process of electron beam induced deposits on various substrates and under varying growth conditions. Apart from the statistical variation in deposited mass as reported earlier [7,8], quite unexpected results are obtained, which are not well understood, yet clearly demonstrate the importance of the substrate and the growth conditions.

Electron Microscopy and Analysis Group Conference 2007 (EMAG 2007) IOP Publishing Journal of Physics: Conference Series 126 (2008) 012025 doi:10.1088/1742-6596/126/1/012025

c

2. Experiment

The EBID experiments were performed in an environmental Scanning Transmission Electron Microscope (STEM) at a beam energy of 200 keV, a 0.2 nm probe and W(CO)6 as a precursor gas.

Typical precursor gas pressures during the deposition are 10-3 Torr, and to reduce contamination from the microscope the substrates are held at an elevated temperature of 100-150 ˚C. Before introducing the precursor gas in the microscope it was always checked whether the contamination level was sufficiently low. Arrays of dots were deposited on thin membranes of i) Si3N4, although the precise

stoichiometry of this material is not really known, and may vary between suppliers, ii) amorphous carbon of several thicknesses, and iii) crystalline BN. The STEM is used in annular dark field (ADF) mode and the ADF signal is measured to monitor the time evolution of the deposits.

3. Results

3.1. Amorphous carbon and crystalline BN substrates

In figure 1a an ADF image is shown of an array of dots deposited on a thin amorphous carbon foil at a temperature of 149 ˚C and at a precursor gas pressure of 4.3·10-3 Torr. Figure 1b shows a similar array of dots deposited at a pressure of 7.4·10-3 Torr and a temperature of 120 ˚C. Although the same mass

Figure 1a. ADF image of an array of dots deposited on an amorphous carbon foil at a W(CO)6 pressure of 4.3·10

-3

Torr and a temperature of 149 ˚C

Figure 1b. ADF image of an array of dots deposited on an amorphous carbon foil at a W(CO)6 pressure of 7.4·10

-3

Torr and a temperature of 120 ˚C

per dot is deposited in both arrays the dots in fig. 1a are much larger than the dots of fig. 1b. Figure 2a shows the average deposited mass per dot for various dwell times under the same growth conditions as

Figure 2a. Average deposited mass per dot versus dwell time. Squares: 7.4·10-3 Torr and 120 ˚C; triangles: 4.3·10-3 Torr and 149 ˚C

Figure 2a. Average dot radius versus average deposited dot mass. Squares: 7.4·10-3 Torr and 120 ˚C; triangles: 4.3·10-3 Torr and 149 ˚C in figure 1. The growth is seen to vary linearly with the dwell time and the lower temperature and higher pressure results in the highest growth rate. In figure 2b the average dot radius (measured at half Electron Microscopy and Analysis Group Conference 2007 (EMAG 2007) IOP Publishing Journal of Physics: Conference Series 126 (2008) 012025 doi:10.1088/1742-6596/126/1/012025

the maximum ADF intensity) is plotted as a function of deposited mass for the two growth conditions, and it is observed that the experiments at the higher temperature and lower pressure always result in the largest dots, even when the same amount of mass is deposited. It is unlikely that the pressure difference of not even a factor of two can explain this result. More likely is that the higher temperature enhances surface effects such as surface diffusion of dissociated precursor fragments. The smaller growth rate at higher temperatures and lower pressure may result from a shorter residence time of the precursor molecules in combination with a lower supply of precursor molecules. Similar results were obtained for arrays of dots deposited on amorphous carbon foils of different thicknesses and on thin flakes of BN crystals.

3.2. Silicon-nitride substrates

When depositing arrays of dots on SiN membranes, well known substrates for TEM experiments, quite surprising results are obtained. Figure 3a shows an array of 5x5 dots deposited on a 10 nm thick SiN

Figure 3a. Array of 5x5 dots deposited on 10 nm thick SiN. Note the 4 unusually bright dots.

Figure 3b. The array of figure 3a tilted over 20 degrees.

Figure 3c. The time dependence of the ADF signal recorded while growing the dots 2, 3, 4, and 5 in figure 3a and b.

membrane, at a dwell time of 8 sec. and a pressure of 2.5·10-3 Torr. The dots seem quite similar in intensity except for 4 dots which are much brighter. When tilting the membrane (Figure 3b) over 20 degrees it is seen that these 4 dots have grown really fast. In figure 3c the time dependence of the ADF signal is plotted for dots 2, 3, 4, and 5 in figure 3a and b. It is observed that initially the growth rate is more or less constant but then suddenly the growth rate increases rapidly for the dots 3 and 5. For these dots the growth rate is also seen to decrease again, which is due to the fact that the ADF signal arises only from the part of the deposit exposed by the primary electron beam, and these fast growing dots are growing out of the primary beam. This is clearly observed in figures 3a and b. This experiment was repeated on a 50 nm thick SiN membrane with similar results, except that even more dots grew exceptionally fast. In Figure 4 the average deposited mass is plotted versus dwell time for

Figure 4. The average deposited mass versus dwell time for arrays of dots deposited at 100 ˚C and a pressure of 2.5·10-3 Torr on 10 nm (open triangles) and 50 nm (solid squares) thick SiN substrates. Electron Microscopy and Analysis Group Conference 2007 (EMAG 2007) IOP Publishing Journal of Physics: Conference Series 126 (2008) 012025 doi:10.1088/1742-6596/126/1/012025

arrays of dots deposited at 100 ˚C and a pressure of 2.5·10-3 Torr on 10 nm and 50 nm thick SiN substrates. The large scatter in the data, especially when compared to the data in figure 2a, is clearly seen. The result of another experiment offers a potential explanation for these observations. In that experiment dots were deposited, at 150 ˚C and 4.3·10-3 Torr and 4 sec. dwell time, on a 50 nm thick SiN membrane with an electrode structure deposited on it, consisting of a bilayer of a few nm of Ti and 10 nm of Ni on top of that. In figure 5a an overview of the membrane with part of the electrode structure is seen. The electrodes are located on the electron beam exit side of the membrane. In figure 5b. an array of dots is shown deposited far away from the electrodes on the SiN. The same unusually

Figure 5. a) SiN membrane with electrode structure on the electron beam exit side, b) array deposited away from the electrodes (position b in fig. 5a), c) array over an electrode (position c in fig. 5a), and d) array close to an electrode just visible in the top of the image (outside of the area of fig. 5a). fast growth as in figure 3 is observed for some of the dots. In figure 5c an array is shown deposited over one of the electrodes and, surprisingly, dots of similar intensity are obtained, comparable to deposits on amorphous carbon. Figure 5d then shows an array deposited very close to one of the electrodes, leading to a regular array as well. This strongly suggests that charging effects play a role [9]. For instance, electric field enhanced surface diffusion of polarized precursor molecules may lead to locally enhanced growth.

4. Conclusion

Experimental results of electron beam induced deposition on amorphous carbon foils and silicon nitride membranes were presented. The diameter of dots deposited on carbon foils was found to depend rather sensitively on the deposition temperature, which is ascribed to diffusion of dissociated precursor molecules. Deposited dots on SiN membranes were shown to display an unusually large variety of heights, which is presumably due to charging effects, leading to locally enhanced growth. The main conclusion from these experiments is that for such small deposits the substrate material and surface condition is of utmost importance to achieve reproducible deposited structures.

References

[1] Liu Z Q, Mitsuishi K and Furuya K 2005 Appl Phys A 80 1437 [2] Guise O, Ahner J, Yates J and Levy J 2004 Appl. Phys. A 85 2352

[3] Jiang H, Borca C N, Xu B and Robertson B W 2001 Int. J. Mod. Phys. B 15 3207 [4] Crozier P A, Tolle J, Kouvetakis J and Ritter C 2004 Appl. Phys. Lett. 84 3441

[5] Van Dorp W F, Van Someren B, Hagen C W, Kruit P and Crozier P A 2005 Nano Lett 5 1303 [6] Silvis-Cividjian N, Hagen C W and Kruit P 2005 J. Appl. Phys. 98 084904

[7] Van Dorp W F, Van Someren B, Hagen C W, Kruit P and Crozier P A 2006 J Vac Sci Technol B 24 618

[8] Van Dorp W F, Hagen C W, Crozier P A, Van Someren B and Kruit P 2006 Microelectron Eng 83 1468

[9] Song M, Mitsuishi K, Tanaka M, Takeguchi M, Shimojo M and Furuya K 2005 Appl. Phys. A 80 1431

Electron Microscopy and Analysis Group Conference 2007 (EMAG 2007) IOP Publishing Journal of Physics: Conference Series 126 (2008) 012025 doi:10.1088/1742-6596/126/1/012025