- Email: [email protected]
AFM lithography of aluminum for fabrication of nanomechanical systems
Ultramicroscopy 97 (2003) 467–472
AFM lithography of aluminum for fabrication of nanomechanical systems Z.J. Davisa,*, G. Abadalb, O. Hansena, X. Borise! b, N. Barniolb, F. Pe! rez-Muranob,c, A. Boisena a Mikroelektronik Centret, Denmark Technical University, Lyngby, DK-2800 Denmark " " Dept. Enginyeria Electronica, Universitat Autonoma Barcelona, E-08193 Bellaterra, Spain c " Institut de Microelectronica Barcelona (IMB-CNM) Campus UAB, 08193 Bellaterra, Spain b
Received 2 July 2002; received in revised form 29 December 2002
Abstract Nanolithography by local anodic oxidation of surfaces using atomic force microscopy (AFM) has proven to be more reproducible when using dynamic, non-contact mode. Hereby, the tip/sample interaction forces are reduced dramatically compared to contact mode, and thus tip wear is greatly reduced. Anodic oxidation of Al can be used for fabricating nanomechanical systems, by using the Al oxide as a highly selective dry etching mask. In our experiments, areas as large as 2 mm 3 mm have been oxidized repeatedly without any sign of tip-wear. Furthermore, line widths down to 10 nm have been routinely obtained, by optimization of AFM parameters, such as tip/sample distance, voltage and scan speed. Finally, AFM oxidation experiments have been performed on CMOS processed chips, demonstrating the first steps of fabricating fully functional nanomechanical devices. r 2003 Elsevier Science B.V. All rights reserved. Keywords: AFM lithography; Non-contact AFM
1. Introduction The need for miniaturization of mechanical systems is being driven by the need for faster, cheaper and more sensitive transducers. To meet the demands of moving from microelectromechanical systems (MEMS) to nanoelectromechanical systems (NEMS), new types of lithography techniques have been exploited. Here we want to *Corresponding author. E-mail address: [email protected] (Z.J. Davis).
highlight AFM lithography, which is inherently a simple, resistless and low cost technique for rapid prototyping. AFM lithography is based on local anodic oxidation of surfaces and has been widely used for the fabrication of nanostructures and nanometer scale devices [1–4]. It has been seen that optimization of the AFM lithography technique, by using non-contact mode AFM, can increase its reliability and reproducibility [5–7]. We will demonstrate that not only the reliability and reproducibility can be improved, but also that this technique is indeed a fast
0304-3991/03/$ - see front matter r 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0304-3991(03)00075-5
468
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Fig. 1. Schematic drawing of the laterally vibrating mass sensor device based on electrostatic excitation and capacitive readout.
nanopatterning prototyping tool and that it is fully compatible with CMOS fabrication technology. Nanomechanical systems can be realized using local oxidation of Al coated Si or SiO2 surfaces [8–10]. In this case the Al is oxidized forming a mask, which, compared to Si or SiO2, is highly selective to dry etching. The mask is developed by a chemical wet etching of the non-oxidized Al. In this communication, the experiments are performed on Al coated Si/SiO2/Si substrates where the SiO2 acts as the sacrificial layer and is a platform for NEMS fabrication. Presently we are developing a mass sensor based on a resonating nanocantilever [11]. The design is very simple and consists of a laterally oscillating cantilever and a parallel electrode. The electrode acts as the driver, when applying an AC voltage between the driver and cantilever (Fig. 1). To realize an integrated capacitive readout of the cantilever’s oscillation, a CMOS chip has been developed on which the nanocantilever can be fabricated as a post-processing module [12] (Fig. 5). The function of the CMOS circuit is to measure and amplify the small capacitive currents that are induced between the cantilever and the driver when the cantilever goes into resonance.
2. Experiments Experiments have been carried out with two different substrates. The first substrate is an Al
coated silicon on insulator (SOI) wafer, with a top Si and SiO2 thickness of 350 and 450 nm, respectively. The top Si layer is first highly doped with boron through ion implantation and subsequently annealed to ensure good electrical conductivity. A boron concentration of approximately 1020 cm 3 is achieved. The second substrate is an Al coated CMOS chip, which has a predefined area with a sandwich layer of poly-Si/SiO2/Si. The thickness of the poly-Si and the SiO2 are 600 nm and 1 mm, respectively. This area is where the mass sensor mechanical device will be fabricated. On both substrates a 7 nm thick Al layer is e-beam deposited prior to the AFM lithography experiments. Nanolithography is performed by using a commercial AFM and lithography software for these non-contact dynamic mode experiments. Before the lithography is performed on the Al surface, an area is scanned and the plane is stored. Then the feedback loop is switched off and the pattern is written on the same area, using the stored plane data to keep the tip/sample distance constant. After the lithography has been performed the feedback loop is activated and the same area is scanned again revealing the oxidized structure. For all the experiments, non-coated ndoped Si tips are used. The cantilever’s resonant frequency is 330 kHz and the vibrational amplitude was approximately 16–20 nm (peak to peak) during scanning. During regular scanning operation the tip/sample distance is approximately 0–2 nm, corresponding to tapping and total noncontact mode.
3. AFM lithography on Al coated SOI Previous works on AFM oxidation lithography on Al coated surfaces have used contact mode AFM [8,9]. In contact mode tip wear is a critical problem because the electrostatic forces between the cantilever and sample are quite large, typically on the order of 100–1000 nN for tip-sample voltages on the order of 10–20 V. This increases the tip/sample interaction dramatically, because the spring constant of contact mode AFM probes is much lower than that of non-contact AFM
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
probes. Furthermore, due to friction between the conducting tip and the Al, which has a thin insulating native oxide layer, electrostatic effects make it difficult to control the tip/sample distance and thus the oxidation is non-reproducible. In non-contact mode lithography these problems are not present. Due to the low interaction forces between tip and sample, tip wear is minimized. Furthermore, since friction is eliminated, the tip/sample distance can be controlled very precisely. Attempts at contact mode AFM imaging of Al coated substrates resulted in the tip hoping up and down during the scan, which is due to the electrostatic effects described above. This tip hoping makes it both difficult to reproducibly oxidize the surface and image the surface afterwards. Furthermore, it was also seen that the life span of the tip increases dramatically from a few hours of writing with contact mode, to several days with non-contact mode lithography. Actually, no sign of tip wear was observed while doing the non-contact AFM lithography experiments presented here. In order to use AFM lithography for the fabrication of NEMS, it is necessary not only to be able to write features with sub-micron dimension, but also large areas, which are needed as connects to contact pads. Therefore, it is advantageous to optimize the lithography for defining small as well as larger features in the same writing process. In non-contact mode lithography, the tipsample distance is easily controlled by monitoring the modulation in the oscillation amplitude of the AFM cantilever [13,14]. Earlier non-contact AFM lithography experiments on Si has shown that a water bridge is formed between the conductive tip and sample and that the geometry of the water bridge influences the final oxide dimensions [5–7]. The formation of the water bridge is observed by monitoring the cantilever’s oscillation amplitude after pulsing with a DC voltage between the tip and sample, while simultaneously disconnecting the feedback loop. After the pulse the oscillation amplitude is lower than the feedback set point, due to the extra forces acting on the tip due to the water bridge. The same phenomenon was observed with the Al coated SOI substrates. During writing the
469
feature size of an oxidized line can be controlled by simply controlling the tip/sample distance and the applied DC voltage. By stretching the water bridge the feature size decreases. By decreasing the tip/sample distance, the water bridge covers a larger area on the Al surface, thus the feature size increases. In contact mode AFM lithography on Al the sudden application of high voltages invokes breakdown, where the surface is destroyed. This breakdown phenomenon is not seen in noncontact mode AFM lithography, thus higher voltages can be applied, which increases even further the obtainable feature sizes. In Fig. 2 both a schematic drawing of this point and two AFM images show how one can vary the feature size from almost 1 mm (Fig. 2a) down to 100 nm (Fig. 2b). In Fig. 2b it is seen that in some of the lines the width is varying along the line. This is explained by the fact that after capturing the plane there was still some drift in the piezoscanner while the lithography is taking place, thus the tip/sample distance is varied along the writing line causing the steady increase or decrease in line width.
Fig. 2. Schematic diagram and non-contact AFM image of an oxidized pattern with (a) a small tip/sample distance and high voltage and (b) large tip/sample distance and low voltage.
470
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Fig. 3. (a) Non-contact AFM image of an oxidized pattern corresponding to the mask needed to make a lateral vibrating cantilever and (b) line profile showing the oxide height is approximately 3.5 nm over the Al, which demonstrates that the Al has been fully oxidized.
By optimizing the AFM writing, large areas can be written in a matter of minutes, when a tip/sample distance of approximately 3–5 nm and a high DC voltage of 30 V or more is chosen. The details of the pattern are written by optimizing the AFM lithography with a larger tip/sample distance of approximately 8–10 nm and a DC voltage of 20 V. The scan speed is usually kept constant at approximately 10 mm/s. Fig. 3a shows an example of a device with two oxidized 2 3 mm2 areas and a very thin, 50 nm wide, line extending from one of the 2 3 mm2 areas. The 50 nm wide line is separated from the other 2 3 mm2 area by a 500 nm gap. Fig. 3b shows a profile of the oxide height showing a height of 3.5 nm, which corresponds to an oxidation of the entire 7 nm Al thickness. In Fig. 4 AFM images of approximately 10 nm Al oxide lines are shown, illustrating that by optimizing the lithography parameters very small feature sizes can be reproducibly obtained. Due to tip convolution the actual width of the oxidized lines could be smaller than 10 nm, but cannot be seen due to the finite tip sharpness of the AFM tip. It can also be seen in the image that the oxide lines vary slightly, which is due to the small sample roughness, which slightly varies the tip/sample distance.
4. AFM lithography on Al coated CMOS In order to demonstrate that the lithography technology can be used for real NEMSs AFM
Fig. 4. Non-contact AFM image of 10 nm wide Al oxide lines on an Al coated SOI substrate.
lithography was performed on a substrate with CMOS circuitry. The whole device including the nanofabrication area and the CMOS circuitry is seen in Fig. 5a. Because the top Si layer is polycrystalline, the roughness is considerately higher than the SOI substrate. This made it very difficult to image the Al oxide mask just after AFM oxidation, but the structures are easily seen with an optical microscope. First, a rough alignment of the AFM tip to the nanofabrication area is done by the optical microscope integrated with the AFM, then the area is scanned and the final
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
471
Fig. 5. Optical images of (a) the entire device consisting of the nanofabrication area and CMOS circuitry, (b) a close-up of the nanofabrication area and (c) a close-up of the Al oxide pattern written by non-contact mode AFM lithography.
alignment is performed. In Fig. 5b a close-up of the nanofabrication area is shown, where the AFM oxidized area is seen as the dark pattern inside the circle. In Fig. 5c, the Al oxide mask is clearly seen. It is also worth noting that the large optical contrast demonstrates that the refractive index of the oxidized Al has changed significantly. The size of the large rectangles is approximately 6 6 mm2 and the width and length of the line and the gap distance between the line and larger square are approximately 500 nm, 8 mm and 500 nm, respectively.
5. Conclusions It has been demonstrated that non-contact AFM lithography has a large potential for the fabrication of NEMSs. Non-contact mode AFM lithography has been shown to have distinct advantages over traditional contact mode AFM
lithography in many ways. First, it has been shown that non-contact AFM lithography is a more robust and reproducible technique, which results in little or no tip wear. Furthermore, it has been shown, due to the dynamic nature of the technique, that the tip/sample distance and the applied voltage can be optimized in a way to achieve both large, micron, feature sizes, and small, nano, feature sizes. This makes this technique highly useful for fast prototyping of NEMS. Finally, mask definition for the fabrication of a mass sensor based on a laterally vibrating nanocantilever has been performed on an SOI and a CMOS chip. These results are the first steps in fabricating a nanocantilever based mass sensor integrated with CMOS circuitry, which will make the sensor more sensitive and with more functionalities than any other sensor of its kind. Complete device fabrication is currently being pursued using AFM defined etch masks as shown in Fig. 3.
472
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Acknowledgements This work is funded by the European IST project NANOMASS II (IST-2001-33068) and is also partially funded by the Microserv program.
References [1] E.S. Snow, D. Park, P.M. Campell, Appl. Phys. Lett. 69 (1996) 269. [2] R. Held, T. Heinzel, P. Studerus, K. ensslin, M. Holland, Appl. Phys. Lett. 71 (1997) 2689. [3] K. Matsumoto, Physica B 227 (1996) 92. [4] F.S.-S. Chein, W.-F. Hsieh, A.E. Vladar, J.A. Dagata, J. Appl. Phys. 91 (2002) 10044. [5] R. Garc!ıa, M. Calleja, F. P!erez-Murano, Appl. Phys. Lett. 72 (1998) 2295.
[6] R. Garc!ıa, M. Calleja, H. Rohrer, J. Appl. Phys. 86 (1999) 1898. [7] M. Calleja, R. Garc!ıa, Appl. Phys. Lett. 76 (2000) 3427. [8] A. Boisen, K. Birkelund, O. Hansen, F. Grey, J. Vac. Sci. Technol. B 16 (1998) 2977. [9] G. Abadal, A. Boisen, Z.J. Davis, O. Hansen, F. Grey, Appl. Phys. Lett. 74 (1999) 3206. [10] G. Abadal, Z.J. Davis, A. Boisen, F. P!erez-Murano, N. Barniol, X. Borris!e, Probe Microsc. 2 (2001) 121. [11] Z.J. Davis, G. Abadal, O. Kuhn, O. Hansen, F. Grey, A. Boisen, J. Vac. Sci. Technol. B 18 (2000) 612. [12] Z.J. Davis, G. Abadal, G. Helbo, O. Hansen, F. Campabadal, F. P!erez-Murano, J. Esteve, E. Figueras, R. Ruiz, N. Barniol, A. Boisen, Proceeding of Transducers ’01, Munich, 2001, pp. 72–75. [13] R. Garc!ıa, A.S. Paulo, Phys. Rev. B 60 (1999) 4961. [14] R. Garc!ıa, A.S. Paulo, Phys. Rev. B 61 (2000) 381.
AFM lithography of aluminum for fabrication of nanomechanical systems Z.J. Davisa,*, G. Abadalb, O. Hansena, X. Borise! b, N. Barniolb, F. Pe! rez-Muranob,c, A. Boisena a Mikroelektronik Centret, Denmark Technical University, Lyngby, DK-2800 Denmark " " Dept. Enginyeria Electronica, Universitat Autonoma Barcelona, E-08193 Bellaterra, Spain c " Institut de Microelectronica Barcelona (IMB-CNM) Campus UAB, 08193 Bellaterra, Spain b
Received 2 July 2002; received in revised form 29 December 2002
Abstract Nanolithography by local anodic oxidation of surfaces using atomic force microscopy (AFM) has proven to be more reproducible when using dynamic, non-contact mode. Hereby, the tip/sample interaction forces are reduced dramatically compared to contact mode, and thus tip wear is greatly reduced. Anodic oxidation of Al can be used for fabricating nanomechanical systems, by using the Al oxide as a highly selective dry etching mask. In our experiments, areas as large as 2 mm 3 mm have been oxidized repeatedly without any sign of tip-wear. Furthermore, line widths down to 10 nm have been routinely obtained, by optimization of AFM parameters, such as tip/sample distance, voltage and scan speed. Finally, AFM oxidation experiments have been performed on CMOS processed chips, demonstrating the first steps of fabricating fully functional nanomechanical devices. r 2003 Elsevier Science B.V. All rights reserved. Keywords: AFM lithography; Non-contact AFM
1. Introduction The need for miniaturization of mechanical systems is being driven by the need for faster, cheaper and more sensitive transducers. To meet the demands of moving from microelectromechanical systems (MEMS) to nanoelectromechanical systems (NEMS), new types of lithography techniques have been exploited. Here we want to *Corresponding author. E-mail address: [email protected] (Z.J. Davis).
highlight AFM lithography, which is inherently a simple, resistless and low cost technique for rapid prototyping. AFM lithography is based on local anodic oxidation of surfaces and has been widely used for the fabrication of nanostructures and nanometer scale devices [1–4]. It has been seen that optimization of the AFM lithography technique, by using non-contact mode AFM, can increase its reliability and reproducibility [5–7]. We will demonstrate that not only the reliability and reproducibility can be improved, but also that this technique is indeed a fast
0304-3991/03/$ - see front matter r 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0304-3991(03)00075-5
468
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Fig. 1. Schematic drawing of the laterally vibrating mass sensor device based on electrostatic excitation and capacitive readout.
nanopatterning prototyping tool and that it is fully compatible with CMOS fabrication technology. Nanomechanical systems can be realized using local oxidation of Al coated Si or SiO2 surfaces [8–10]. In this case the Al is oxidized forming a mask, which, compared to Si or SiO2, is highly selective to dry etching. The mask is developed by a chemical wet etching of the non-oxidized Al. In this communication, the experiments are performed on Al coated Si/SiO2/Si substrates where the SiO2 acts as the sacrificial layer and is a platform for NEMS fabrication. Presently we are developing a mass sensor based on a resonating nanocantilever [11]. The design is very simple and consists of a laterally oscillating cantilever and a parallel electrode. The electrode acts as the driver, when applying an AC voltage between the driver and cantilever (Fig. 1). To realize an integrated capacitive readout of the cantilever’s oscillation, a CMOS chip has been developed on which the nanocantilever can be fabricated as a post-processing module [12] (Fig. 5). The function of the CMOS circuit is to measure and amplify the small capacitive currents that are induced between the cantilever and the driver when the cantilever goes into resonance.
2. Experiments Experiments have been carried out with two different substrates. The first substrate is an Al
coated silicon on insulator (SOI) wafer, with a top Si and SiO2 thickness of 350 and 450 nm, respectively. The top Si layer is first highly doped with boron through ion implantation and subsequently annealed to ensure good electrical conductivity. A boron concentration of approximately 1020 cm 3 is achieved. The second substrate is an Al coated CMOS chip, which has a predefined area with a sandwich layer of poly-Si/SiO2/Si. The thickness of the poly-Si and the SiO2 are 600 nm and 1 mm, respectively. This area is where the mass sensor mechanical device will be fabricated. On both substrates a 7 nm thick Al layer is e-beam deposited prior to the AFM lithography experiments. Nanolithography is performed by using a commercial AFM and lithography software for these non-contact dynamic mode experiments. Before the lithography is performed on the Al surface, an area is scanned and the plane is stored. Then the feedback loop is switched off and the pattern is written on the same area, using the stored plane data to keep the tip/sample distance constant. After the lithography has been performed the feedback loop is activated and the same area is scanned again revealing the oxidized structure. For all the experiments, non-coated ndoped Si tips are used. The cantilever’s resonant frequency is 330 kHz and the vibrational amplitude was approximately 16–20 nm (peak to peak) during scanning. During regular scanning operation the tip/sample distance is approximately 0–2 nm, corresponding to tapping and total noncontact mode.
3. AFM lithography on Al coated SOI Previous works on AFM oxidation lithography on Al coated surfaces have used contact mode AFM [8,9]. In contact mode tip wear is a critical problem because the electrostatic forces between the cantilever and sample are quite large, typically on the order of 100–1000 nN for tip-sample voltages on the order of 10–20 V. This increases the tip/sample interaction dramatically, because the spring constant of contact mode AFM probes is much lower than that of non-contact AFM
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
probes. Furthermore, due to friction between the conducting tip and the Al, which has a thin insulating native oxide layer, electrostatic effects make it difficult to control the tip/sample distance and thus the oxidation is non-reproducible. In non-contact mode lithography these problems are not present. Due to the low interaction forces between tip and sample, tip wear is minimized. Furthermore, since friction is eliminated, the tip/sample distance can be controlled very precisely. Attempts at contact mode AFM imaging of Al coated substrates resulted in the tip hoping up and down during the scan, which is due to the electrostatic effects described above. This tip hoping makes it both difficult to reproducibly oxidize the surface and image the surface afterwards. Furthermore, it was also seen that the life span of the tip increases dramatically from a few hours of writing with contact mode, to several days with non-contact mode lithography. Actually, no sign of tip wear was observed while doing the non-contact AFM lithography experiments presented here. In order to use AFM lithography for the fabrication of NEMS, it is necessary not only to be able to write features with sub-micron dimension, but also large areas, which are needed as connects to contact pads. Therefore, it is advantageous to optimize the lithography for defining small as well as larger features in the same writing process. In non-contact mode lithography, the tipsample distance is easily controlled by monitoring the modulation in the oscillation amplitude of the AFM cantilever [13,14]. Earlier non-contact AFM lithography experiments on Si has shown that a water bridge is formed between the conductive tip and sample and that the geometry of the water bridge influences the final oxide dimensions [5–7]. The formation of the water bridge is observed by monitoring the cantilever’s oscillation amplitude after pulsing with a DC voltage between the tip and sample, while simultaneously disconnecting the feedback loop. After the pulse the oscillation amplitude is lower than the feedback set point, due to the extra forces acting on the tip due to the water bridge. The same phenomenon was observed with the Al coated SOI substrates. During writing the
469
feature size of an oxidized line can be controlled by simply controlling the tip/sample distance and the applied DC voltage. By stretching the water bridge the feature size decreases. By decreasing the tip/sample distance, the water bridge covers a larger area on the Al surface, thus the feature size increases. In contact mode AFM lithography on Al the sudden application of high voltages invokes breakdown, where the surface is destroyed. This breakdown phenomenon is not seen in noncontact mode AFM lithography, thus higher voltages can be applied, which increases even further the obtainable feature sizes. In Fig. 2 both a schematic drawing of this point and two AFM images show how one can vary the feature size from almost 1 mm (Fig. 2a) down to 100 nm (Fig. 2b). In Fig. 2b it is seen that in some of the lines the width is varying along the line. This is explained by the fact that after capturing the plane there was still some drift in the piezoscanner while the lithography is taking place, thus the tip/sample distance is varied along the writing line causing the steady increase or decrease in line width.
Fig. 2. Schematic diagram and non-contact AFM image of an oxidized pattern with (a) a small tip/sample distance and high voltage and (b) large tip/sample distance and low voltage.
470
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Fig. 3. (a) Non-contact AFM image of an oxidized pattern corresponding to the mask needed to make a lateral vibrating cantilever and (b) line profile showing the oxide height is approximately 3.5 nm over the Al, which demonstrates that the Al has been fully oxidized.
By optimizing the AFM writing, large areas can be written in a matter of minutes, when a tip/sample distance of approximately 3–5 nm and a high DC voltage of 30 V or more is chosen. The details of the pattern are written by optimizing the AFM lithography with a larger tip/sample distance of approximately 8–10 nm and a DC voltage of 20 V. The scan speed is usually kept constant at approximately 10 mm/s. Fig. 3a shows an example of a device with two oxidized 2 3 mm2 areas and a very thin, 50 nm wide, line extending from one of the 2 3 mm2 areas. The 50 nm wide line is separated from the other 2 3 mm2 area by a 500 nm gap. Fig. 3b shows a profile of the oxide height showing a height of 3.5 nm, which corresponds to an oxidation of the entire 7 nm Al thickness. In Fig. 4 AFM images of approximately 10 nm Al oxide lines are shown, illustrating that by optimizing the lithography parameters very small feature sizes can be reproducibly obtained. Due to tip convolution the actual width of the oxidized lines could be smaller than 10 nm, but cannot be seen due to the finite tip sharpness of the AFM tip. It can also be seen in the image that the oxide lines vary slightly, which is due to the small sample roughness, which slightly varies the tip/sample distance.
4. AFM lithography on Al coated CMOS In order to demonstrate that the lithography technology can be used for real NEMSs AFM
Fig. 4. Non-contact AFM image of 10 nm wide Al oxide lines on an Al coated SOI substrate.
lithography was performed on a substrate with CMOS circuitry. The whole device including the nanofabrication area and the CMOS circuitry is seen in Fig. 5a. Because the top Si layer is polycrystalline, the roughness is considerately higher than the SOI substrate. This made it very difficult to image the Al oxide mask just after AFM oxidation, but the structures are easily seen with an optical microscope. First, a rough alignment of the AFM tip to the nanofabrication area is done by the optical microscope integrated with the AFM, then the area is scanned and the final
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
471
Fig. 5. Optical images of (a) the entire device consisting of the nanofabrication area and CMOS circuitry, (b) a close-up of the nanofabrication area and (c) a close-up of the Al oxide pattern written by non-contact mode AFM lithography.
alignment is performed. In Fig. 5b a close-up of the nanofabrication area is shown, where the AFM oxidized area is seen as the dark pattern inside the circle. In Fig. 5c, the Al oxide mask is clearly seen. It is also worth noting that the large optical contrast demonstrates that the refractive index of the oxidized Al has changed significantly. The size of the large rectangles is approximately 6 6 mm2 and the width and length of the line and the gap distance between the line and larger square are approximately 500 nm, 8 mm and 500 nm, respectively.
5. Conclusions It has been demonstrated that non-contact AFM lithography has a large potential for the fabrication of NEMSs. Non-contact mode AFM lithography has been shown to have distinct advantages over traditional contact mode AFM
lithography in many ways. First, it has been shown that non-contact AFM lithography is a more robust and reproducible technique, which results in little or no tip wear. Furthermore, it has been shown, due to the dynamic nature of the technique, that the tip/sample distance and the applied voltage can be optimized in a way to achieve both large, micron, feature sizes, and small, nano, feature sizes. This makes this technique highly useful for fast prototyping of NEMS. Finally, mask definition for the fabrication of a mass sensor based on a laterally vibrating nanocantilever has been performed on an SOI and a CMOS chip. These results are the first steps in fabricating a nanocantilever based mass sensor integrated with CMOS circuitry, which will make the sensor more sensitive and with more functionalities than any other sensor of its kind. Complete device fabrication is currently being pursued using AFM defined etch masks as shown in Fig. 3.
472
Z.J. Davis et al. / Ultramicroscopy 97 (2003) 467–472
Acknowledgements This work is funded by the European IST project NANOMASS II (IST-2001-33068) and is also partially funded by the Microserv program.
References [1] E.S. Snow, D. Park, P.M. Campell, Appl. Phys. Lett. 69 (1996) 269. [2] R. Held, T. Heinzel, P. Studerus, K. ensslin, M. Holland, Appl. Phys. Lett. 71 (1997) 2689. [3] K. Matsumoto, Physica B 227 (1996) 92. [4] F.S.-S. Chein, W.-F. Hsieh, A.E. Vladar, J.A. Dagata, J. Appl. Phys. 91 (2002) 10044. [5] R. Garc!ıa, M. Calleja, F. P!erez-Murano, Appl. Phys. Lett. 72 (1998) 2295.
[6] R. Garc!ıa, M. Calleja, H. Rohrer, J. Appl. Phys. 86 (1999) 1898. [7] M. Calleja, R. Garc!ıa, Appl. Phys. Lett. 76 (2000) 3427. [8] A. Boisen, K. Birkelund, O. Hansen, F. Grey, J. Vac. Sci. Technol. B 16 (1998) 2977. [9] G. Abadal, A. Boisen, Z.J. Davis, O. Hansen, F. Grey, Appl. Phys. Lett. 74 (1999) 3206. [10] G. Abadal, Z.J. Davis, A. Boisen, F. P!erez-Murano, N. Barniol, X. Borris!e, Probe Microsc. 2 (2001) 121. [11] Z.J. Davis, G. Abadal, O. Kuhn, O. Hansen, F. Grey, A. Boisen, J. Vac. Sci. Technol. B 18 (2000) 612. [12] Z.J. Davis, G. Abadal, G. Helbo, O. Hansen, F. Campabadal, F. P!erez-Murano, J. Esteve, E. Figueras, R. Ruiz, N. Barniol, A. Boisen, Proceeding of Transducers ’01, Munich, 2001, pp. 72–75. [13] R. Garc!ıa, A.S. Paulo, Phys. Rev. B 60 (1999) 4961. [14] R. Garc!ıa, A.S. Paulo, Phys. Rev. B 61 (2000) 381.