Development of Crosstalk Eliminated (XE) AFM

Introduction

The Atomic Force Microscope (AFM) is a powerful instrument for nano-meter scale science and technology. Since its invention in 1986¹, the AFM has evolved by refining its capabilities and ease-of-use. The most common configuration uses a micro-machined cantilever with a sharp tip on its edge which scans over a sample utilizing a piezoelectric tube scanner (Figure 1). The deflection of the cantilever is measured by casting a laser beam onto the cantilever’s backside and then detecting the reflected beam with a position sensitive photo detector (PSPD) ².

Figure 1. Piezoelectric tube scanner used in conventional AFM. When a mirror symmetric voltage is applied to the opposite electrodes, the tube bends side ways. However, it is not an orthogonal 3D actuator.

In such a configuration, an AFM has high vertical sensitivity and is relatively easy to implement. In order to adjust the incident laser beam to fall on the small cantilever and to make the reflected beam hit the center of the PSPD, an aligning mechanism with fine screws is used. This probing assembly unit, including an aligning mechanism, a laser, a PSPD, and cantilever, has a considerable mass and it is difficult to scan the probing unit at a sufficiently high speed while accurately measuring a sample. Therefore in earlier AFMs, the probing unit was kept stationary, and the sample was scanned along the XY and Z axes as shown in Figure 2.

However, such an AFM has an intrinsic problem of the Z-servo performance being dependent on the sample mass. Since the sample has to be moved in the Z direction, the Z-servo control parameters need to be adjusted every time the sample mass changes. A more serious problem arises when it is necessary to image large samples such as large silicon wafers, which cannot be scanned fast enough in the Z direction for sufficient vertical servo frequency response. In order to solve this problem, it is necessary to scan the probe cantilever and to ensure the laser beam follows the cantilever’s motion.

A simple means of accomplishing this result is to miniaturize the aligning mechanism and to scan the entire probing unit as shown in Figure 3.³ However, such a miniaturized probing unit still has a considerable mass which degrades the Z-servo response. It is also inconvenient to align the laser beam with tiny screws and a special tool is often required.

Another method is to attach lenses on the tube scanner such that the laser beam follows the cantilever motion and the reflected beam hits the same point on the PSPD while scanning, as shown in Figure 4.^4,5 However, in this method, the laser beam does not perfectly follow the cantilever motion and the reflected beam does not remain on the exact same point on the PSPD, causing measurement errors and tracking force variations during XY scanning.

Figure 2. In most AFMs, a laser beam bounce system is used to detect the cantilever deflection and a piezoelectric tube scanner is used to scan the sample in XY and z directions. Large samples cannot be scanned.

Figure 3. For large samples, the tube scanner may scan the entire miniaturized probing unit. However, it is inconvenient to align the laser beam, and the Z-servo response is slow due to the increased mass that must be scanned.

Figure 4. Another method is to insert lenses inside the tube scanner to make the laser beam follow the cantilever motion. However, this method is complicated and there are residual errors in the laser beam path.

In addition, most AFMs have problems with scanning errors and slow scanning speeds. As shown in Figure 1, the commonly used piezoelectric tube scanner is not an orthogonal 3 dimensional actuator that can be moved in any of the three dimensions x, y, and z independently of one another. Since the XY motion relies on the bending of the tube, there is non-linearity and serious cross talk between the XY and Z axes.

Position sensors can be used to correct the intrinsic non-linearity of the piezoelectric actuator,⁶ but the cross talk from flexing the tube cannot be eliminated and it causes background curvature and measurement errors. Using a tripod scanner does not improve the non-linearity and cross talk problem much. Furthermore, the tube scanner has a low resonance frequency (typically below 500 kHz) and does not have enough force to drive a conventional probing unit at high speed.

Advanced XE Scan System

Park Systems’ advanced Crosstalk Elimination (XE) scan system (shown in Figure 5) effectively addresses all of the above-mentioned problems. In this configuration, we used a 2-dimentional flexure stage to scan the sample in only the XY direction, and a stacked piezoelectric actuator to scan the probe cantilever in the Z direction only. The flexure stage used for the XY scanner is made of solid aluminum as shown in Figure 6. It demonstrates high orthogonality and an excellent out-of-plane motion profile. The flexure stage can scan large samples (~1 kg) up to a few 100 Hz in the XY direction. This scan speed is sufficient because the bandwidth requirement for the XY axes is much lower than that for the Z axis. The stacked piezoelectric actuator for the Z-scanner has a high resonance frequency (~10 kHz) with a high pushpull force when appropriately pre-loaded.


Figure 5. In XE systems, the Z-scanner is separated from the XY- scanner; the XY-scanner scans only the sample and the Z-scanner scans only the probe.		Figure 6. XY flexure scanner used in XE-system. This single module parallel-kinematics stage has low inertia and minimal out-of-plane motion, providing the best orthogonality, high responsiveness, and axis-independent performance.

A crucial point of our design is the arrangement of the laser, PSPD, and the laser beam aligning mechanism. As mentioned above, we have to ensure that the laser beam falls on the same point on the cantilever and the reflected beam hits the same point on the PSPD regardless of the Z-scanner motion, so that only the deflection of the cantilever will be monitored on the PSPD. We can achieve this goal by casting the laser beam vertically from above and attaching the PSPD to the Z-scanner, while the laser and laser beam aligning mechanism are fixed to the frame. As shown in Figure 7 (a), the laser is mounted on one side of the probing head. The laser beam is reflected by a prism, which is mounted on a glass plate. The angle of the glass plate can be adjusted by the two screws on the two diagonal corners of the glass plate holder. Since the laser beam is falling on the cantilever from the vertical direction, the beam always hits the same point on the cantilever regardless of the Z-scanner motion. The reflected beam is bounced at the steering mirror and hits the PSPD. The angle of the steering mirror can be slightly adjusted by the two screws on its diagonal edges such that the bounced beam hits the center of the PSPD. Since the PSPD and cantilever move together and the steering mirror is vertically mounted, parallel with the Z-scan direction, the bounced laser beam always hits the same point on the PSPD regardless of the Z-scanner motion.

Figure 7. (a) Initial design of the beam bounce detection mechanism for the XE scan system. To detect the cantilever deflection, the PSPD and the cantilever are moved together by the Z-scanner, while the laser, laser aligning mechanism, and steering mirror are fixed to the head frame. (b) One variation of the beam bounce detection mechanism. The PSPD was lowered to make clearance for the optical microscope. However, small errors are introduced in this design. (c) Another variation of the beam bounce detection mechanism. In order to eliminate the error, a second mirror that is parallel to the steering mirror was inserted. This mirror exactly compensates for the effect of the steering mirror.

In order to accommodate an on-axis optical microscope, it is desirable to have suitable clearance above the cantilever. For this purpose, we lowered the position of the PSPD and mounted the steering mirror at a certain angle such that the path of the bounced laser beam became horizontal, as shown in Figure 7 (b). However, in this configuration, the bounced laser beam spot on the PSPD changes as the Z-scanner moves. When the Z-scanner moves a distance h, there is an error of h(1-sin2θ) in the position of the laser beam spot on the PSPD, where θ is the angle of the cantilever. This error term is very small compared to the amount of the laser beam spot displacement when the cantilever is deflected by a feature of height h on the sample surface. However, this phenomenon still causes height measurement errors and introduces spurious variations in the tracking force. Since the Z-scanner motion is a known quantity, we can compensate for such errors by software.

A better method is to eliminate such errors by introducing another mirror, whose angle is in parallel with the steering mirror, and positioning the PSPD accordingly, as shown in Figure 7 (c). In this configuration, the second mirror exactly compensates for the effect of the first mirror, and therefore the laser beam hits the same point on the PSPD regardless of the Z-scanner motion.

Figure 8. (a)Optical microscope view and (b)sectional diagram of the XE AFM stage This design allows direct on-axis optical view.

This design provides clearance above the cantilever and allows for a direct on-axis optical microscope view as shown in Figure 8. The optical path from the sample to the camera is a straight line. This configuration provides much higher quality optical views than was provided by conventional large sample AFMs,^3,4 where an oblique mirror had to be inserted between the cantilever and the objective lens. Since the oblique mirror may have some defects and as it does not fully cover the light path, the quality of the optical microscope is degraded in these systems. Also, in order to pan the view, the objective lens has to be moved out of its optical axis, introducing significant blurring. In the XE-system, the objective lens, tube lens, and CCD camera are rigidly mounted on a single body and move together for panning and focusing to preserve the highest quality optical vision.

XE Performance

XE-series AFMs are not only the most technologically-advanced AFMs on the market, but also the most user-friendly. The intuitive user interface minimizes the time needed to start taking measurements and reflects our dedication to user-oriented product development. XE-series AFMs allow customers of all experience levels to focus on their experiments, not instruments. From developing the first direct on-axis optics in industrial systems to our simplified tip exchange procedure, Park Systems has led trend-setting and market leading innovation in user convenience. At Park Systems, the ease of use is the ultimate support we can provide to our valued customers.

Figure 9. Zero background curvature by Park Systems XE-system (a) and typical background curvature of a conventional AFM system with a tube scanner (b). (c) shows the cross section of these background curvatures.

Figure 9 shows unprocessed AFM images of a bare silicon wafer taken with the XE-system (a), and with a conventional AFM7 (b). Since the silicon wafer is atomically flat, most of the curvatures in the image are scanner-induced artifacts. Figure 9 (c) shows the cross section of the images in Figure 9 (a) and (b). Since the tube scanner has intrinsic background curvatures, the maximum out-of-plane motion is as much as 80 nm when the X-axis moves 15 μm. The XE scan system has less than 1nm of out-of-plane motion for the same scan range. Another advantage of the XE scan system is its Z-servo response. Figure 10 is an image of a porous polymer sphere (Styrene Divinyl Benzene), whose diameter is about 5 µm, taken with the XE-system in Non-Contact mode. Since the Z-servo response of the XE-system is very accurate, the probe can precisely follow the steep curvature of the polymer sphere as well as small porous surface structures without crashing or sticking to the surface. Figure 11 shows another example that demonstrates the high performance of the z-servo response with a flat background.

Figure 10. NC-AFM image of a polymer sphere taken with an
XE-100 (6 μm scan size). 1:1 aspect ratio of un-processed raw data..

Figure 11. NC-AFM image of STI patterns on a 6 inch photomask taken with an XE- 150 (5 μm × 5μm scan, 70 nm z range). 3D rendering of un-processed raw data. The AFM probe traced both upper and lower terraces faithfully and almost imaged the side walls.

Conclusion

The new XE-series SPM was developed to have the following advantages: 1) Scan accuracy: There is no cross talk between the XY and Z axes, and one can achieve high scan accuracy. 2) Sample size: Since the sample is scanned by a flexure scanner only in the XY direction, large samples as well as small samples can be scanned at sufficiently high speed. 3) Z-servo speed: Since the Z-scanner has a high resonance frequency with high force, the Z-servo frequency response is much greater than in conventional AFM. 4) Convenience: Since the laser beam aligning mechanism is fixed to the head, it is possible to produce alignment fixtures of adequate size for convenient and precise adjustment without requiring any tools. 5) Optical vision: Since there is enough clearance above the cantilever, it is possible to accommodate a direct on-axis optical microscope.

References

1 G. Binnig, C. Quate, and Ch. Gerber, Phys. Rev. Lett. 56, 930 (1986).
2 G. Meyer and N. M. Amer, Appl. Phys. Lett. 53, 2400 (1988).
3 Agilent (formerly Molecular Imaging). Model 5500 (formerly PicoPlus)
4 Veeco (formerly Digital Instruments), Model Dimension 3100
5 P. K. Hansma, B. Drake, D. Grigg, C. B. Prater, F. Yashar, G. Gurley, and V. Elings, S. Feinstein, and R. Lal, J. Appl. Phys. 76, 796 (1994).
6 R. Barrett, Rev. Sci. Instrum. 62, 1393 (1991).
7 Veeco (formerly Digital Instruments), Model Multimode