- Email: [email protected]
Conductive transparent fiber probes for shear-force atomic force microscopes
ARTICLE IN PRESS
Ultramicroscopy 106 (2006) 146–151 www.elsevier.com/locate/ultramic
Conductive transparent fiber probes for shear-force atomic force microscopes Tooru Murashita Nippon Telegraph and Telephone Corporation, NTT Photonics Laboratories, 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan Received 10 September 2004; received in revised form 18 May 2005; accepted 22 June 2005
Abstract New conductive transparent (CT) probes that can inject currents into nanometer-sized regions and collect light from them have been developed for shear-force atomic force microscopy (SF-AFM) of partially isolative regions. The CT probe consists of a straight elastic silica fiber with one end tapered to a point. The taper is coated with an indium-tinoxide film as a transparent electrode, and the probe apex has a nanometer-scale radius. The essential feature of the CT probes is coaxial nickel plating on the shaft of the isolative silica fiber, which is adjusted to obtain suitable elasticity for smooth shear-force feedback as well as for supplying currents to the transparent electrode. Experimental results clarified that nickel thickness between 0.5 and 15 mm on 20 mm-long fibers makes resistance low enough for supplying current to the probe apex and also makes the Q curves smooth enough for shear-force feedback. Clear SF-AFM and current images were successfully obtained for a sample containing both conductive and isolative regions. The CT probes for SF-AFM can expand applications of probe-current-induced luminescence measurements to samples that contain highly resistive and isolative regions, for which scanning tunneling microscopy cannot be applied. r 2005 Elsevier B.V. All rights reserved. Keywords: Tunneling electron; Luminescence; Probe; AFM; Metal plating; Optical fiber
1. Introduction Real-space characterizations of the electronic and optical properties of structures and materials in local nanometer-sized regions are essential for nanotechnologies, which deal with nanometersized structures and materials, such as quantum Tel.: +462402598; fax: +462402872.
E-mail address: [email protected].
devices and individual molecules. One way to obtain such information is to use tunnelingelectron induced luminescence (TL) microscopes. The TL microscope injects tunneling electrons into a nanometer-sized region of a sample and collects the resulting luminescence. They usually use combinations of a metal probe for injecting currents and a lens or mirror for collecting the luminescence. The lenses or mirrors, however, have low-collection yields because their solid
0304-3991/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ultramic.2005.06.061
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
angles are small and they can not collect optical nearfields. To improve the collection yield, we have already developed a conductive transparent (CT) probe [1,2]. The apex of the probe injects electrons into a nanometer-sized local region of a sample and simultaneously collects electron-induced luminescence from the region. The CT probe consists of a straight multimode silica fiber with a core diameter of 100 mm for applying spectroscopy in wide wavelength ranges. One end of the fiber tapers to a point having a nanometerscale radius. The tapered part is coated with indium-tin-oxide (ITO) film to a thickness of less than several ten nanometers to make a transparent electrode. The straight probe produces quite low transmission loss of the light passing through it. The key features of this probe are that the apex simultaneously injects electrons into a nanometersized region of a sample and collects the resulting luminescence. Since the probe is placed within just a few nanometers of the sample, the collection yield for luminescence and even for evanescent light, is much greater than that when lenses or mirrors are used. Using the CT probe, we have successfully performed real-space electronic and optical characterizations of highly doped semiconductors. For instance, we obtained real-space TL images of the cross section of GaAs/AlAs multiple quantum wells with spatial resolution of less than 10 nm, where each layer was 50 nm thick [3]. However, samples sometimes contain highly resistive or isolative regions. The CT probe developed for scanning tunneling microscopy (STM) is not suitable for such samples because STM requires enough probe current at every point in the scanned area to control the gap between the probe and sample. Consequently, we need a new type of CT probe that can control the gap without using a probe current. A good solution is to use atomic force, which is independent of the sample conductivity. Atomic forces can be detected by frequency shifts of the probe vibrations close to the sample. There are mainly four methods of controlling the gap in atomic force microscopy (AFM) operations, and they all use some combination of probe vibration direction and a method of probe
147
vibration detection. The two probe-vibration directions are longitudinal (tapping mode) and parallel (shear-force mode) to the sample surface. The variation in the dithering amplitude is detected optically by using an optical lever with a laser beam or electrically by using a tuning fork. The CT probe should be straight for the low-loss transmission of light. A straight probe is better suited for operation in the shear force mode than in the tapping mode, which requires a cantilever [4,5]. Moreover, a laser beam should not be used in order to avoid its influence on weak TL measurements in nanometer-sized regions. Consequently, we selected a combination of shear-force-mode AFM (SF-AFM) and a tuning fork. The probe of the SF-AFM vibrates parallel to the sample surface. The CT probes for SF-AFM need suitable elasticity in order to produce smooth vibration and obtain sufficient sensitivity for atomic force detection. The CT probe we previously developed for STM is short and thickly plated with nickel to make it stiff and thereby reduce noise from mechanical vibrations. The STM-CT probe is not suitable for SF-AFM, because a SF-AFM probe should be moderately elastic so that it can faithfully follow oscillations. In addition to sufficient elasticity, it must have good conductivity and transmissivity as well. We have therefore developed new CT probes for SF-AFM. In this article, we describe the structures and basic performance of the novel CT probe for SF-AFM.
2. Structure of the CT probe for SF-AFM Fig. 1 shows the fundamental setup of our SFAFM with the CT probe. We use a commercially available SF-AFM with a quartz tuning fork mounted on a quartz dithering oscillator. The distance between the holding plate and the fulcrum is about 17 mm and is determined by the configuration of the apparatus. Taking into account the margin needed for mounting, we made the probe 20 mm long. The CT probe is fixed on the metal holder of the SF-AFM with the metal holding plate, which is connected to a bias supply through a current limiter. The tuning fork touches the
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
148 light holding plate
current limiter
thick metal plating
multimode silica fiber coaxial metal plating
probe length (20 mm)
electric isolator bias tuning oscillator fork
ITO coated taper apex probe current sample
Fig. 1. Structure of the CT probe for shear-force AFM.
probe and acts as a fulcrum at a point about 3 mm from the apex. The fiber was bent a little to create a pressure on the tuning fork. If any current leaks from the probe into the tuning fork, the tuning fork will not operate properly. Therefore, the tuning fork is insulated from the CT probe. The quartz oscillator drives the probe, and the sample is adjusted in the vertical direction so as to maintain a constant atomic force. The amplitude of the dithering oscillation parallel to the sample surface is about 20 nm at the probe apex in free space. The straight CT probe for SF-AFM consists of a multimode silica fiber, which is the same fiber used for the STM-CT probe and has a core diameter of 100 mm and an outer diameter of 140 mm. One end of the fiber is tapered to a point. The taper is coated with a thin indium-tin-oxide (ITO) film of about 70 nm thickness to make a transparent electrode. The radius of the ITOcoated apex of the probe is less than 100 nm. The transmittance and sheet resistivity of the ITO film are over 70% and 100 Ocm, respectively. Compared to metals, ITO films are also advantageous because they allow robust high bias voltages of up to 10 V and large probe currents of several hundred nano-amperes compared to metals. Nickel was coaxially plated on the shaft of the fiber, expect for the taper part coated with ITO. The CT probe is fastened to a base mount with the
metal holding plate, which works as an electric terminal that supplies current from the bias supply to the CT probe. The part where the probe is held was thickly plated to a thickness of more than 600 mm (outer diameter) so that it would be rigid enough to withstand a large holding pressure. The tuning fork touches the CT probe as a fulcrum positioned about 17 mm from the edge of the metal holding plate and about 3 mm from the front end of the CT probe. The tuning fork is sensitive to electrical noise, so it and the CT probe are electrically isolated. The frequency of the dithering oscillation with the tuning fork is 30.43 kHz. The direction of the dithering oscillation is parallel to the sample surface and normal to the probe scanning direction. The sample is mounted on a piezo-driven stage. The shear force for adjusting the gap between the probe and sample is monitored as the variation in amplitude of the probe vibration at the monitoring frequency. With the above setup, three things need to be considered in designing CT probes for SF-AFM operations. The first is that the natural frequency of the probe should be different from the oscillation frequency of the quartz oscillator, which is around 30 kHz for our SF-AFM. The second is that the resistance should be low. The third is that the probe should have a high mechanical toughness. Nickel was selected for the fiber plating because it adheres well to silica. Young’s modulus of nickel (200 GPa) is much larger than that of silica (74 GPa) so that only several micrometers of plating can mostly determine the elasticity of the probe, which is essential for good SF-AFM performance. Although electroplating alone provides sufficient mechanical toughness, the nickel could be directly plated on the silica fibers by nonelectric plating. However, the thickness of nonelectric nickel plating is less than 0.5 mm, which is not suitable for practical use because the plating is mechanically weak and highly resistive. For practical use as the probe, therefore, nickel of more than 0.5 mm thickness is plated by electroplating on non-electroplating, and nickel can be plated to a thickness greater than 0.5 mm for probe fabrication.
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
Fig. 3. Dependence of resistance of the nickel plating on plating thickness.
free running
with atomic force
0.25 vibration amplitude (a.u.)
The elasticity can be adjusted by changing the thickness of the metal plating on the fiber. Rigidity approximately increases as thickness increases. When the thickness is greater than 15 mm, plastic deformation occurs, which is the main factor determining the upper limit of plating thickness. Fig. 2 shows the calculated resonant frequency of the probe as a function of the thickness of the nickel coaxial plating on the fiber [6]. Although the resonant frequency increases as plating thickness increases, the frequency does not exceed 11 kHz for plating thickness between 0.5 and 15 mm, which is far enough from the oscillation frequency (around 30 kHz) for stable probe operation. The resistance and elasticity of the CT probe were measured for nickel-plating thickness of between 0.5 and 10 mm. For electroplated nickel, the resistivity was 1 O/cm2 and constant, so that the resistance is inversely proportional to a crosssection of the coaxial nickel plating as shown in Fig. 3. The resistance for these thicknesses is less than about 60 O over the full length of the probe (20 mm), which is low enough for probe current supply. Consequently, nickel-plating thickness between 0.5 and 15 mm is applicable for CT probes in terms of smooth performance of SF-AFM and probe current supply.
149
feedback 0.20
0.15
0.10
sensing 30660
30700
driving 30740
30780
frequency (Hz) 12 resonant frequency (kHz)
Fig. 4. Q curves of the CT probe. 11 10
3. Characteristics of the AFM with the CT probe
9 8 7
outer diameter of fiber: 140 µm
6 0
5
10
15
coaxial nickel plate thickness (µm)
Fig. 2. Resonant frequencies as a function of nickel plating thickness.
Fig. 4 shows the typical resonance frequency spectrum, or Q curve, of the CT probe with 4 mmthick nickel plating. The Q curve with a higher peak frequency indicates the free-running state far from a sample, and that with lower peak frequency is measured at the distance at which the atomic force is detected. The peak of the Q curve for the free-running state is at 30.74 kHz. The Q curve around the peak is smooth enough for stable feedback control. Since the Q curve is asymmetric,
ARTICLE IN PRESS 150
T. Murashita / Ultramicroscopy 106 (2006) 146–151
the lower-frequency-side slope is used for feedback control. In this case, we set the monitoring frequency to 30.73 kHz. When the probe detects atomic forces, the peak of the Q curve shifts toward lower frequencies, and the vibration amplitude at the monitoring frequency also changes. The feedback controller adjusts the feedback gain so as to maintain the amplitude of the vibration of the probe at the monitoring frequency. The distance between the probe and the sample is adjusted so as to maintain feedback values in the amplitude of the vibration of the probe at the monitoring frequency. The CT probes stably approached both conductive and isolative regions on samples as follows. In the case of an isolative sample consisting of 2 mm square chromium patterns on an isolative glass substrate, the CT probes stably approached and obtained SF-AFM images of the patterns. Fig. 5 shows then the probe current and amplitude of the lateral probe vibration as a function of the distance between the probe and a gold-coated conductive sample. The distance at which the probe current begins to increase is shorter than that where the shear-force mode atomic force begins to decrease. Thus, the probe current is much more sensitive to the distance than the shear-force mode atomic force. In this case, probe current maximized at the completion of the approach. This indicates that the probe may have slightly contacted the sample. Therefore, the AFM
operates in the contact mode and probe currents were thus saturated while the probe vibration amplitude was minimized. For TL measurements, the bias usually ranges between 1 and 10 V, in which most of the energy levels of materials exist, and the probe current reaches several ten nanoamperes. These values are much higher than those for ordinary STM measurements. We investigated whether or not a SF-AFM with the CT probe could operate stably under such measurement conditions, even when the resistance of the sample changed. The sample was germanium antimony tellurium (GST) deposited on a conductive silicon wafer. The resistivities are markedly different between the resistive amorphous and conductive crystalline phases of GST. Fig. 6 shows the probe current passing through a 20 nm-thick GST layer as a function of bias. In this case, the GST was initially in the amorphous phase. At a low bias, the GST was resistive amorphous and the probe current was as low as the detection limit for STM operations. At a high bias of over 8.5 V, the GST changed to the conductive crystalline phase because of the probecurrent heating. Then, the conductivity increased markedly by three orders of magnitude. Thus, the SF-AFM CT probe can reliably measure the probe current in materials in which the conductivity changes by three orders. Fig. 7 shows simultaneously taken SF-AFM and current images of a 20 mm-square area of a sample on which there was an about 1 mm f small isolative phase change amorphous phase crystalline phase (resistive) (conductive) probe current (nA)
10 1 0.1 0.0 0.001 0 Fig. 5. Probe current and amplitude of the lateral probe vibration as a function of the distance between the probe and sample.
CT probe GST Pt Sub. detection limit (STM unstable) 2
4 6 probe bias (V)
8
10
Fig. 6. I/V characteristics of phase change material measured using the CT probe.
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
(a)
(b) probe
isolative particle ITO
Fig. 7. Atomic force (a) and probe current (b) images of a small isolative particle on a conductive ITO film.
particle on a conductive ITO film sputtered on a glass substrate. The ITO film stably maintains its conductivity in air. The ITO film was connected to an electrode of the bias supply with conductive silver paste. The isolative small particle on the ITO film (indicated by arrows) appears distinctly as a dark spot in the current image. These experimental results clarify that the CT probes successfully perform both SF-AFM operations and probe current injections on samples containing both conductive and isolative regions. Although the probe vibration amplitude determines the spatial resolution of the SF-AFM image, which was 20 nm for this equipment, the resolution of both the current and current-induced luminescence images can be improved by loading the probe bias only at the center of the vibration, regardless of the probe vibration amplitude.
4. Conclusion We have developed a new CT probe for SFAFM of samples that contain both conductive and
151
isolative regions. We found that the proper elasticity for SF-AFM operations was obtained with the appropriate nickel-plating thickness between 0.5 and 15 mm for a 20 mm-long straight probe. A SF-AFM with the CT probe operated stably and produced good SF-AFM and current images for samples containing both conductive and isolative regions. Next, we are planning to do optical measurements, with the ultimate goal of conducting realspace measurements of the electronic and optical properties of nanometer-sized areas even when the sample contains both conductive and isolative regions.
Acknowledgement We thank Takatomo Enoki and Takashi Kobayashi of NTT Photonics Laboratories for their continuous encouragement, and also thank Masato Iyoki of SII Nanotechnology Inc. for his fruitful suggestions and support in the experiments.
References [1] [2] [3] [4]
T. Murashita, J. Vac. Sci. Technol. B 15 (1997) 32. T. Murashita, J. Electron Microsc. 46 (1997) 199. T. Murashita, K. Tateno, Appl. Phys. Lett. 78 (2001) 3995. E. Betzig, P.L. Finn, J.S. Weiner, Appl. Phys. Lett. 60 (1995) 2484. [5] M. Vaez-Iravani, R. Toledo-Crow, Y. Chen, J. Vac. Sci. Technol. A 11 (1993) 742. [6] T.R. Albrecht, P. Grutter, D. Horme, D. Rugar, J. Appl. Phys. 69 (1991) 668.
Ultramicroscopy 106 (2006) 146–151 www.elsevier.com/locate/ultramic
Conductive transparent fiber probes for shear-force atomic force microscopes Tooru Murashita Nippon Telegraph and Telephone Corporation, NTT Photonics Laboratories, 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan Received 10 September 2004; received in revised form 18 May 2005; accepted 22 June 2005
Abstract New conductive transparent (CT) probes that can inject currents into nanometer-sized regions and collect light from them have been developed for shear-force atomic force microscopy (SF-AFM) of partially isolative regions. The CT probe consists of a straight elastic silica fiber with one end tapered to a point. The taper is coated with an indium-tinoxide film as a transparent electrode, and the probe apex has a nanometer-scale radius. The essential feature of the CT probes is coaxial nickel plating on the shaft of the isolative silica fiber, which is adjusted to obtain suitable elasticity for smooth shear-force feedback as well as for supplying currents to the transparent electrode. Experimental results clarified that nickel thickness between 0.5 and 15 mm on 20 mm-long fibers makes resistance low enough for supplying current to the probe apex and also makes the Q curves smooth enough for shear-force feedback. Clear SF-AFM and current images were successfully obtained for a sample containing both conductive and isolative regions. The CT probes for SF-AFM can expand applications of probe-current-induced luminescence measurements to samples that contain highly resistive and isolative regions, for which scanning tunneling microscopy cannot be applied. r 2005 Elsevier B.V. All rights reserved. Keywords: Tunneling electron; Luminescence; Probe; AFM; Metal plating; Optical fiber
1. Introduction Real-space characterizations of the electronic and optical properties of structures and materials in local nanometer-sized regions are essential for nanotechnologies, which deal with nanometersized structures and materials, such as quantum Tel.: +462402598; fax: +462402872.
E-mail address: [email protected].
devices and individual molecules. One way to obtain such information is to use tunnelingelectron induced luminescence (TL) microscopes. The TL microscope injects tunneling electrons into a nanometer-sized region of a sample and collects the resulting luminescence. They usually use combinations of a metal probe for injecting currents and a lens or mirror for collecting the luminescence. The lenses or mirrors, however, have low-collection yields because their solid
0304-3991/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ultramic.2005.06.061
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
angles are small and they can not collect optical nearfields. To improve the collection yield, we have already developed a conductive transparent (CT) probe [1,2]. The apex of the probe injects electrons into a nanometer-sized local region of a sample and simultaneously collects electron-induced luminescence from the region. The CT probe consists of a straight multimode silica fiber with a core diameter of 100 mm for applying spectroscopy in wide wavelength ranges. One end of the fiber tapers to a point having a nanometerscale radius. The tapered part is coated with indium-tin-oxide (ITO) film to a thickness of less than several ten nanometers to make a transparent electrode. The straight probe produces quite low transmission loss of the light passing through it. The key features of this probe are that the apex simultaneously injects electrons into a nanometersized region of a sample and collects the resulting luminescence. Since the probe is placed within just a few nanometers of the sample, the collection yield for luminescence and even for evanescent light, is much greater than that when lenses or mirrors are used. Using the CT probe, we have successfully performed real-space electronic and optical characterizations of highly doped semiconductors. For instance, we obtained real-space TL images of the cross section of GaAs/AlAs multiple quantum wells with spatial resolution of less than 10 nm, where each layer was 50 nm thick [3]. However, samples sometimes contain highly resistive or isolative regions. The CT probe developed for scanning tunneling microscopy (STM) is not suitable for such samples because STM requires enough probe current at every point in the scanned area to control the gap between the probe and sample. Consequently, we need a new type of CT probe that can control the gap without using a probe current. A good solution is to use atomic force, which is independent of the sample conductivity. Atomic forces can be detected by frequency shifts of the probe vibrations close to the sample. There are mainly four methods of controlling the gap in atomic force microscopy (AFM) operations, and they all use some combination of probe vibration direction and a method of probe
147
vibration detection. The two probe-vibration directions are longitudinal (tapping mode) and parallel (shear-force mode) to the sample surface. The variation in the dithering amplitude is detected optically by using an optical lever with a laser beam or electrically by using a tuning fork. The CT probe should be straight for the low-loss transmission of light. A straight probe is better suited for operation in the shear force mode than in the tapping mode, which requires a cantilever [4,5]. Moreover, a laser beam should not be used in order to avoid its influence on weak TL measurements in nanometer-sized regions. Consequently, we selected a combination of shear-force-mode AFM (SF-AFM) and a tuning fork. The probe of the SF-AFM vibrates parallel to the sample surface. The CT probes for SF-AFM need suitable elasticity in order to produce smooth vibration and obtain sufficient sensitivity for atomic force detection. The CT probe we previously developed for STM is short and thickly plated with nickel to make it stiff and thereby reduce noise from mechanical vibrations. The STM-CT probe is not suitable for SF-AFM, because a SF-AFM probe should be moderately elastic so that it can faithfully follow oscillations. In addition to sufficient elasticity, it must have good conductivity and transmissivity as well. We have therefore developed new CT probes for SF-AFM. In this article, we describe the structures and basic performance of the novel CT probe for SF-AFM.
2. Structure of the CT probe for SF-AFM Fig. 1 shows the fundamental setup of our SFAFM with the CT probe. We use a commercially available SF-AFM with a quartz tuning fork mounted on a quartz dithering oscillator. The distance between the holding plate and the fulcrum is about 17 mm and is determined by the configuration of the apparatus. Taking into account the margin needed for mounting, we made the probe 20 mm long. The CT probe is fixed on the metal holder of the SF-AFM with the metal holding plate, which is connected to a bias supply through a current limiter. The tuning fork touches the
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
148 light holding plate
current limiter
thick metal plating
multimode silica fiber coaxial metal plating
probe length (20 mm)
electric isolator bias tuning oscillator fork
ITO coated taper apex probe current sample
Fig. 1. Structure of the CT probe for shear-force AFM.
probe and acts as a fulcrum at a point about 3 mm from the apex. The fiber was bent a little to create a pressure on the tuning fork. If any current leaks from the probe into the tuning fork, the tuning fork will not operate properly. Therefore, the tuning fork is insulated from the CT probe. The quartz oscillator drives the probe, and the sample is adjusted in the vertical direction so as to maintain a constant atomic force. The amplitude of the dithering oscillation parallel to the sample surface is about 20 nm at the probe apex in free space. The straight CT probe for SF-AFM consists of a multimode silica fiber, which is the same fiber used for the STM-CT probe and has a core diameter of 100 mm and an outer diameter of 140 mm. One end of the fiber is tapered to a point. The taper is coated with a thin indium-tin-oxide (ITO) film of about 70 nm thickness to make a transparent electrode. The radius of the ITOcoated apex of the probe is less than 100 nm. The transmittance and sheet resistivity of the ITO film are over 70% and 100 Ocm, respectively. Compared to metals, ITO films are also advantageous because they allow robust high bias voltages of up to 10 V and large probe currents of several hundred nano-amperes compared to metals. Nickel was coaxially plated on the shaft of the fiber, expect for the taper part coated with ITO. The CT probe is fastened to a base mount with the
metal holding plate, which works as an electric terminal that supplies current from the bias supply to the CT probe. The part where the probe is held was thickly plated to a thickness of more than 600 mm (outer diameter) so that it would be rigid enough to withstand a large holding pressure. The tuning fork touches the CT probe as a fulcrum positioned about 17 mm from the edge of the metal holding plate and about 3 mm from the front end of the CT probe. The tuning fork is sensitive to electrical noise, so it and the CT probe are electrically isolated. The frequency of the dithering oscillation with the tuning fork is 30.43 kHz. The direction of the dithering oscillation is parallel to the sample surface and normal to the probe scanning direction. The sample is mounted on a piezo-driven stage. The shear force for adjusting the gap between the probe and sample is monitored as the variation in amplitude of the probe vibration at the monitoring frequency. With the above setup, three things need to be considered in designing CT probes for SF-AFM operations. The first is that the natural frequency of the probe should be different from the oscillation frequency of the quartz oscillator, which is around 30 kHz for our SF-AFM. The second is that the resistance should be low. The third is that the probe should have a high mechanical toughness. Nickel was selected for the fiber plating because it adheres well to silica. Young’s modulus of nickel (200 GPa) is much larger than that of silica (74 GPa) so that only several micrometers of plating can mostly determine the elasticity of the probe, which is essential for good SF-AFM performance. Although electroplating alone provides sufficient mechanical toughness, the nickel could be directly plated on the silica fibers by nonelectric plating. However, the thickness of nonelectric nickel plating is less than 0.5 mm, which is not suitable for practical use because the plating is mechanically weak and highly resistive. For practical use as the probe, therefore, nickel of more than 0.5 mm thickness is plated by electroplating on non-electroplating, and nickel can be plated to a thickness greater than 0.5 mm for probe fabrication.
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
Fig. 3. Dependence of resistance of the nickel plating on plating thickness.
free running
with atomic force
0.25 vibration amplitude (a.u.)
The elasticity can be adjusted by changing the thickness of the metal plating on the fiber. Rigidity approximately increases as thickness increases. When the thickness is greater than 15 mm, plastic deformation occurs, which is the main factor determining the upper limit of plating thickness. Fig. 2 shows the calculated resonant frequency of the probe as a function of the thickness of the nickel coaxial plating on the fiber [6]. Although the resonant frequency increases as plating thickness increases, the frequency does not exceed 11 kHz for plating thickness between 0.5 and 15 mm, which is far enough from the oscillation frequency (around 30 kHz) for stable probe operation. The resistance and elasticity of the CT probe were measured for nickel-plating thickness of between 0.5 and 10 mm. For electroplated nickel, the resistivity was 1 O/cm2 and constant, so that the resistance is inversely proportional to a crosssection of the coaxial nickel plating as shown in Fig. 3. The resistance for these thicknesses is less than about 60 O over the full length of the probe (20 mm), which is low enough for probe current supply. Consequently, nickel-plating thickness between 0.5 and 15 mm is applicable for CT probes in terms of smooth performance of SF-AFM and probe current supply.
149
feedback 0.20
0.15
0.10
sensing 30660
30700
driving 30740
30780
frequency (Hz) 12 resonant frequency (kHz)
Fig. 4. Q curves of the CT probe. 11 10
3. Characteristics of the AFM with the CT probe
9 8 7
outer diameter of fiber: 140 µm
6 0
5
10
15
coaxial nickel plate thickness (µm)
Fig. 2. Resonant frequencies as a function of nickel plating thickness.
Fig. 4 shows the typical resonance frequency spectrum, or Q curve, of the CT probe with 4 mmthick nickel plating. The Q curve with a higher peak frequency indicates the free-running state far from a sample, and that with lower peak frequency is measured at the distance at which the atomic force is detected. The peak of the Q curve for the free-running state is at 30.74 kHz. The Q curve around the peak is smooth enough for stable feedback control. Since the Q curve is asymmetric,
ARTICLE IN PRESS 150
T. Murashita / Ultramicroscopy 106 (2006) 146–151
the lower-frequency-side slope is used for feedback control. In this case, we set the monitoring frequency to 30.73 kHz. When the probe detects atomic forces, the peak of the Q curve shifts toward lower frequencies, and the vibration amplitude at the monitoring frequency also changes. The feedback controller adjusts the feedback gain so as to maintain the amplitude of the vibration of the probe at the monitoring frequency. The distance between the probe and the sample is adjusted so as to maintain feedback values in the amplitude of the vibration of the probe at the monitoring frequency. The CT probes stably approached both conductive and isolative regions on samples as follows. In the case of an isolative sample consisting of 2 mm square chromium patterns on an isolative glass substrate, the CT probes stably approached and obtained SF-AFM images of the patterns. Fig. 5 shows then the probe current and amplitude of the lateral probe vibration as a function of the distance between the probe and a gold-coated conductive sample. The distance at which the probe current begins to increase is shorter than that where the shear-force mode atomic force begins to decrease. Thus, the probe current is much more sensitive to the distance than the shear-force mode atomic force. In this case, probe current maximized at the completion of the approach. This indicates that the probe may have slightly contacted the sample. Therefore, the AFM
operates in the contact mode and probe currents were thus saturated while the probe vibration amplitude was minimized. For TL measurements, the bias usually ranges between 1 and 10 V, in which most of the energy levels of materials exist, and the probe current reaches several ten nanoamperes. These values are much higher than those for ordinary STM measurements. We investigated whether or not a SF-AFM with the CT probe could operate stably under such measurement conditions, even when the resistance of the sample changed. The sample was germanium antimony tellurium (GST) deposited on a conductive silicon wafer. The resistivities are markedly different between the resistive amorphous and conductive crystalline phases of GST. Fig. 6 shows the probe current passing through a 20 nm-thick GST layer as a function of bias. In this case, the GST was initially in the amorphous phase. At a low bias, the GST was resistive amorphous and the probe current was as low as the detection limit for STM operations. At a high bias of over 8.5 V, the GST changed to the conductive crystalline phase because of the probecurrent heating. Then, the conductivity increased markedly by three orders of magnitude. Thus, the SF-AFM CT probe can reliably measure the probe current in materials in which the conductivity changes by three orders. Fig. 7 shows simultaneously taken SF-AFM and current images of a 20 mm-square area of a sample on which there was an about 1 mm f small isolative phase change amorphous phase crystalline phase (resistive) (conductive) probe current (nA)
10 1 0.1 0.0 0.001 0 Fig. 5. Probe current and amplitude of the lateral probe vibration as a function of the distance between the probe and sample.
CT probe GST Pt Sub. detection limit (STM unstable) 2
4 6 probe bias (V)
8
10
Fig. 6. I/V characteristics of phase change material measured using the CT probe.
ARTICLE IN PRESS T. Murashita / Ultramicroscopy 106 (2006) 146–151
(a)
(b) probe
isolative particle ITO
Fig. 7. Atomic force (a) and probe current (b) images of a small isolative particle on a conductive ITO film.
particle on a conductive ITO film sputtered on a glass substrate. The ITO film stably maintains its conductivity in air. The ITO film was connected to an electrode of the bias supply with conductive silver paste. The isolative small particle on the ITO film (indicated by arrows) appears distinctly as a dark spot in the current image. These experimental results clarify that the CT probes successfully perform both SF-AFM operations and probe current injections on samples containing both conductive and isolative regions. Although the probe vibration amplitude determines the spatial resolution of the SF-AFM image, which was 20 nm for this equipment, the resolution of both the current and current-induced luminescence images can be improved by loading the probe bias only at the center of the vibration, regardless of the probe vibration amplitude.
4. Conclusion We have developed a new CT probe for SFAFM of samples that contain both conductive and
151
isolative regions. We found that the proper elasticity for SF-AFM operations was obtained with the appropriate nickel-plating thickness between 0.5 and 15 mm for a 20 mm-long straight probe. A SF-AFM with the CT probe operated stably and produced good SF-AFM and current images for samples containing both conductive and isolative regions. Next, we are planning to do optical measurements, with the ultimate goal of conducting realspace measurements of the electronic and optical properties of nanometer-sized areas even when the sample contains both conductive and isolative regions.
Acknowledgement We thank Takatomo Enoki and Takashi Kobayashi of NTT Photonics Laboratories for their continuous encouragement, and also thank Masato Iyoki of SII Nanotechnology Inc. for his fruitful suggestions and support in the experiments.
References [1] [2] [3] [4]
T. Murashita, J. Vac. Sci. Technol. B 15 (1997) 32. T. Murashita, J. Electron Microsc. 46 (1997) 199. T. Murashita, K. Tateno, Appl. Phys. Lett. 78 (2001) 3995. E. Betzig, P.L. Finn, J.S. Weiner, Appl. Phys. Lett. 60 (1995) 2484. [5] M. Vaez-Iravani, R. Toledo-Crow, Y. Chen, J. Vac. Sci. Technol. A 11 (1993) 742. [6] T.R. Albrecht, P. Grutter, D. Horme, D. Rugar, J. Appl. Phys. 69 (1991) 668.