Nanometer scale lithography of silicon(100) surfaces using tapping mode atomic force microscopy J. Servat, a) P. Gorostiza, and F. Sanz Department Química-Fisica, Universitat de Barcelona, 08028 Barcelona, Spain F. Pérez-Murano, N. Barniol, G. Abadal, and X. Aymerich b) Department Física-Electrònica, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain Received 2 October 1995; accepted 18 March 1996 Si 100 surfaces have been successfully oxidized at nanometer scale using an atomic force microscope working in tapping mode TMAFM. To modify the surface, gold coated tips and chromium gold coated tips have been used in order to apply a positive voltage to the sample against the grounded tip. A silicon oxide line of 10 nm lateral dimensions can be routinely grown on Si 100 surfaces by TMAFM, at a tip velocity as high as 0.1 mm/s. Pattern dimensions have been measured for different tip velocities and applied voltages and a tip velocity of up to 10 mm/s has been predicted. The patterns have been successfully used as a lithographic mask for a wet chemical etching. 1996 American Vacuum Society. I. INTRODUCTION Local oxidation of silicon surfaces by scanning tunneling microscopy STM and atomic force microscopy AFM is today one of the most promising techniques to modify surfaces at nanometer scale. Since the first work by Dagata et al., 1 the main application of this modification technique has been nanolithography; the oxide is used as a mask for dry 2,3 or wet 3 9 etching processes in the fabrication of several kinds of nanostructures: metallic nanowires, 10 patterning of metal oxide semiconductor field effect transistors MOSFETs 11 and side-gate FETs, 12 selective deposition of gold. 13 Nevertheless, the physicochemical mechanism is not yet well known, although it is quite clear that local silicon oxidation is a field induced reaction 7,14 in which the ambient humidity may play an important role. 15 Tapping mode AFM TMAFM 16 has become a powerful technique that combines some of the advantages of contact and noncontact AFM. In TMAFM, the cantilever is forced to oscillate with a high amplitude tens of nanometers near its resonance frequency around 300 khz, so that the tip taps the surface in every oscillation cycle. Thus TMAFM minimizes both lateral forces and the interaction time with the sample. The force exerted to the sample during TMAFM imaging can be reduced to lower values than in contact mode 17 but at the same TMAFM resolution is higher than in noncontact AFM. Regarding to the patterning of silicon surfaces, both contact and noncontact AFM have been used. Although patterns have been written with high resolution 10 nm and velocity in contact mode, 7 either the surface or the tip can be easily damaged during imaging or patterning. These problems can be overcome in noncontact mode 18 but longer times are necessary to perform a modification. In this article we show that using TMAFM, high resolution and velocity patterning on a Electronic mail: jordis@littlefly.ffn.ub.es b Electronic mail: ifel1@cc.uab.es silicon can be produced reducing the damage on both tip and sample. In a previous contribution, 19 the fabrication of oxide dots on silicon surfaces with 10 nm of lateral dimensions was reported. Here, we perform oxide lines 10 nm wide, and we achieve writing velocities of 0.1 mm/s. Such oxide patterns were successfully transferred to the silicon by wet chemical etching, as other groups reported, 3 9 showing that TMAFM is a fine tool for nanometer scale lithography. The influence of TMAFM parameters i.e., oscillation amplitude and resonance frequency on the oxidation process is discussed. II. EXPERIMENT All the experiments have been performed with a Nanoscope III operating in air. 20 During the experiments, the laboratory has registered humidity levels ranging between 60% and 80%. Commercially available cantilevers 21 have been metallized in order to increase their conductivity. Two kinds of coatings have been tested: a evaporation of 30 nm of gold and b evaporation of 10 nm of chromium plus 30 nm of gold. It was found that both coatings can be employed to oxidize the surface, although tips a sometimes produce gold depositions on the surface as has been reported before, 22 which are almost prevented by using tips b. As the original tips are microfabricated from n-type silicon surface resistivity 0.01 0.02 cm, we have also succeeded in oxidizing the silicon surface with uncoated tips, but in this case the lifetime of the tip drastically decreases. The samples were cut from a highly doped n-type Si 100 wafer surface resistivity: 0.02 cm and it is important to denote that oxide growth can be induced on both native oxide covered surfaces and HF passivated surfaces. In the first case, the sample does not need any cleaning procedure. To produce a modification, the cantilever oscillation amplitude which is used as the feedback signal in TMAFM must be set to a relatively low value around 5 nm. The feedback loop is disabled while the voltage is applied to the 1208 J. Vac. Sci. Technol. A 14(3), May/Jun 1996 0734-2101/96/14(3)/1208/5/$10.00 1996 American Vacuum Society 1208
1209 Servat et al.: Oxidation of Si(100) using TMAFM 1209 FIG. 1. TMAFM image of a native oxide covered Si 100 surface in which 100 dots 30 nm wide and 1 nm high have been written. To write each dot, a voltage pulse of 11 V and 10 ms is applied between the sample and the tip. sample with respect to the grounded tip using the Nanoscope built-in voltage source. A reduction of the amplitude is then observed due to the electrostatic force between tip and surface. Two kinds of modification, a dot and a line, have been performed. To write a dot, the feedback loop is disabled and a voltage pulse is applied while the tip is oscillating at a fixed point on the surface. To write a line, the feedback loop is disabled, the voltage is switched on and the tip is scanned on the surface at a fixed velocity. During the line scan, the tip oscillates at a constant height. Once the line is finished, the voltage is switched off and the feedback enabled. III. RESULTS FIG. 2. TMAFM image of a H-passivated Si 100 surface in which three oxide lines, 1 m long, have been written applying a voltage of 11 V. Starting at the thinnest line, 10 nm wide, the velocity used to write the lines is 100, 10 and 1 m/s, respectively. In order to demonstrate the reliability of the oxidation process, 100 dots were written in a square micrometer of a silicon sample with a native oxide layer Fig. 1. The spacing between two single dots is 100 nm. Pulses of 11 V and 10 ms were applied to create a lattice of reproducible dots 1 nm high and 30 nm wide. It took 1.5 s to create the whole pattern. Patterning of lines was also routinely performed. The oxidation parameters were optimized to write lines of minimum width at the maximum velocity. Figure 2 is a TMAFM image showing three oxide lines grown on a H-passivated silicon sample at 11 V and three different velocities. The thinnest line is 10 nm wide and was written at a velocity of 0.1 mm/s. We have tested the quality of the grown oxide as a mask for nanometer scale lithography by selectively etching the patterned sample. Prior to patterning, the sample was passivated in a HF solution HF 40%:H 2 O 1:10 during 15 s and rinsed in low conductivity water. Several 1 m long oxide lines were written and then immersed in 6.3 M KOH aqueous solution at 60 C for 1.5 s. After rinsing in low conductivity water and drying with argon, the sample was imaged using TMAFM. The same coated tip was used for both modification and imaging. Figures 3 a amd 3 b show five lines before and after selective etching, respectively. The oxide lines in Fig. 3 a were written at 11 V and a velocity of 0.5 mm/s. Lines are 1 m long, 2.5 nm high, 100 nm wide and spaced 140, 70, and 15 nm from top to bottom. The two last lines are overlapped. Figure 3 b shows the same area after KOH etching, with the pattern transferred to the substrate. The resulting lines are 130 nm wide and 19 nm high and the spacing between them is now 109 nm for the first spacing and 23 nm for the second. The minimum separation of 23 nm is also important for lithography, and it could be reduced if the etching parameters were optimized. Given a tip, the line dimensions can be controlled varying the tip velocity or the applied voltage. Two experiments were performed. Figure 4 a shows five lines written at a fixed tip velocity 0.5 mm/s and voltages varying from 12 to 8 V. A linear dependence of the oxide dimensions with respect to the voltage is observed. Figure 4 b shows the same area after the KOH etching. In Fig. 4 c the voltage is kept at 11 V and the tip velocities are 10, 1, 0.5, 0.1, and 0.05 m/s. A logarithmic dependence of the line dimensions with the tip velocity is obtained. The same area after KOH etching is shown in Fig. 4 d. This result is not surprising given the dependence of the dot dimensions with the voltage value and pulse duration. 19 IV. DISCUSSION The oxidation process can be explained as a field induced anodization, where the high electric field of the tip produces oxyanions that recombine with holes at the Si surface to create Si O bonds. 7 Snow et al., measured a threshold time assigned to this reaction from the shortest square wave voltage pulse necessary to modify the surface. This threshold time depends exponentially on the voltage, with measured JVST A - Vacuum, Surfaces, and Films
1210 Servat et al.: Oxidation of Si(100) using TMAFM 1210 FIG. 3. a TMAFM image of a H-passivated Si 100 surface in which five oxide lines, 1 m long, 2.5 nm high and 100 nm wide, have been induced. To write each line, the tip is scanned over the surface at a velocity of 0.5 m/s, and with a bias voltage of 11 V. b TMAFM image of the same lines after a wet etching in KOH. The height of the resulting lines is now 19 nm. values ranging between 1 and 10 4 s for voltage values ranging between 4 and 12 V, in the case of n-type silicon. In our previous paper 19 we concluded that in TMAFM, despite the maximum electric field is the same different oscillation amplitudes of the cantilever, the threshold voltage of the anodization process does depend on this oscillation amplitude. Thus, the electric field must be higher than a threshold value for a sufficient period of the oscillation cycle in order to modify the silicon surface. This dependence is in agreement with the concept of a threshold time in the sense that anodization requires a longer time than the tip oscillation period a few milliseconds. In other words, for a given tip a threshold oscillation amplitude exists above which the modification is not induced. Since the feedback is disabled during the application of the voltage pulse, the oscillation amplitude is forced to de- J. Vac. Sci. Technol. A, Vol. 14, No. 3, May/Jun 1996
1211 Servat et al.: Oxidation of Si(100) using TMAFM 1211 FIG. 4. a TMAFM image of five oxide lines written on a H-passivated Si 100 surface. The written velocity of the cantilever is 0.5 m/s and the voltage applied for each line is from top to bottom : 12, 11, 10, 9, and 8 V. b TMAFM image of the same five lines imaged in a after a KOH etching. c TMAFM image of five oxide lines written on H-passivated Si 100 surface. To write the lines, the applied voltage is 11 V and the scan velocity of the cantilever for each line is from top to bottom : 10, 1, 0.5, 0.1, and 0.05 m/s. d TMAFM image of the same five lines as in c after a KOH etching. crease. So, we have to distinguish between the initial oscillation amplitude with the feedback loop enabled and without applying any voltage and the final oscillation amplitude with the feedback disabled and with a voltage applied between tip and surface. We have found that the value of the final oscillation amplitude has to be very low around 0.1 nm, smaller than the thickness of a water monolayer covering the surface. It suggests that the water layer may play an important role in the modification process, in agreement with the observed dependence on the air humidity reported by others authors. 15 Then, as the oscillation amplitude is so low, the conditions during the oxidation are almost the same for the major part of the oscillation cycle close proximity between tip and surface which gives place to a high electrical field, as it had been found before. 19 In lithography, the maximum velocity of pattern writing is very important. As shown in Fig. 2 we can raise velocities of 0.1 mm/s, that is the highest velocity reported, to the best of our knowledge, in silicon anodization. In order to estimate the maximum pattern velocity expected for the oxidation process, the line height of the modification measured in Fig. 4 c has been plotted against the logarithm of the velocity Fig. 5, showing a linear dependence. An extrapolation of the data to zero height which means no modification by assuming a linear behavior, gives an estimation of the maximum velocity which is over 1 mm/s open squares, a closed value to that predicted by Snow et al. 7 For different tips, the plot of the oxide height as a function of the velocity gives linear dependencies with the same slope and different maximum value, that is, a fine tip could produce written rates up to 10 mm/s. In particular, we have tried to write a line over the maximum predicted velocity and no modification was observed. V. CONCLUSIONS Tapping mode AFM has been successfully used to locally oxidize silicon surfaces performing dots and lines. TMAFM has the advantage of combining permanent interaction during modification like in contact AFM and minimum damage during imaging like noncontact AFM. Also, we have shown that the TMAFM oxidation is a very reliable, high resolution 10 nm and high speed process 0.1 mm/s. Fine tips would JVST A - Vacuum, Surfaces, and Films
1212 Servat et al.: Oxidation of Si(100) using TMAFM 1212 FIG. 5. Plot of the line height measured on the modification against the logarithm of the scan velocity of the cantilever. Both lines have been performed applying a voltage of 11 V with two different tips. permit reaching pattern velocities up to 10 mm/s. The grown oxide has been demonstrated to resist a wet chemical etching, enabling TMAFM to be a powerful tool for nanometer scale lithography on silicon surfaces. ACKNOWLEDGMENTS The authors are indebted to Esteve Farrés from the CNM for metallization of the cantilevers. They would also like to acknowledge the Serveis Científico-Tècnics of the University of Barcelona for equipment availability. This work was supported by the DGICYT through Project No. PB-92-0587, CICYT through Project No. MAT94-1338, and Generalitat de Catalunya through Project No. XT94-15. 1 J. A. Dagata, J. Schneir, H. H. Harary, C. J. Evans, M. T. Postek, and J. Bennett, Appl. Phys. Lett. 56, 2001 1990. 2 P. Fay, R. T. Laughed, G. Abeln, P. Scott, S. Agarwala, I. Adesida, and J. W. Lyding, J. Appl. Phys. 75, 7545 1994. 3 E. S. Snow, W. H. Juan, S. W. Pang, and P. M. Campbell, Appl. Phys. Lett. 66, 1729 1995. 4 E. S. Snow, P. M. Campbell, and B. V. Shanabrook, Appl. Phys. Lett. 63, 3488 1993. 5 S. C. Minne, H. T. Soh, Ph. Flueckiger, and C. F. Quate, J. Vac. Sci. Technol. B 13, 1380 1995. 6 N. Kramer, H. Birk, J. Jorritsma, and C. Schönenberger, Microelectron. Eng. 27, 47 1995. 7 E. S. Snow and P. M. Campbell, Appl. Phys. Lett. 64, 1932 1994. 8 H. Sugimura and N. Nakagiri, Jpn. J. Appl. Phys. 34, 3406 1995. 9 H. Sugimura, T. Yamamoto, N. Nakagiri, M. Miyashita, and T. Onuki, Appl. Phys. Lett. 65, 1569 1994. 10 N. Kramer, H. Birk, J. Jorritsma, and C. Schönenberger, Appl. Phys. Lett. 66, 1325 1994. 11 S. C. Minne, H. T. Soh, Ph. Flueckiger, and C. F. Quate, Appl. Phys. Lett. 64, 703 1995. 12 P. M. Campbell, E. S. Snow, and P. J. McMarr, Appl. Phys. Lett. 66, 1388 1995. 13 H. Sugimura and N. Nakagiri, Appl. Phys. Lett. 66, 1430 1995. 14 F. Pérez-Murano, N. Barniol, and X. Aymerich, J. Vac. Sci. Technol. B 11, 651 1993. 15 H. Sugimira, T. Uchida, N. Kitamura, and H. Masuhara, Appl. Phys. Lett. 63, 1288 1993. 16 Q. Zhog, D. Inniss, K. Kjoller, and V. B. Ellings, Surf. Sci. Lett. 290, L688 1993. 17 J. P. Spatz, S. Sheiko, M. Möller, R. G. Wnikler, P. Reineker, and O. Marti, Nanotechnology 6, 40 1995. 18 D. Wang, L. Tsau, and K. L. Wang, Appl. Phys. Lett. 65, 1415 1994. 19 F. Pérez-Murano, G. Abadal, N. Barniol, X. Aymerich, J. Servat, P. Gorostiza, and F. Sanz, J. Appl. Phys. 78, 6797 1995. 20 Digital Instruments, Santa Barbara, CA. 21 Nanosensors, Germany. 22 R. Imura, H. Koyanagi, M. Miyamoto, A. Kikukawa, T. Shintani, and S. Hogaka, Microelectron. Eng. 27, 105 1995. J. Vac. Sci. Technol. A, Vol. 14, No. 3, May/Jun 1996