Ultrashort cantilever probes for high-speed atomic force microscopy

Since its invention in 1986, atomic force microscopy (AFM) has become the most widely used scanning-probe imaging technique.¹ The microscope ‘maps’ the topography of the sample by scanning its surface with a small tip integrated near the free end of a flexible cantilever. In contrast to other microscope techniques, it is possible to image the surface of many materials—including electrically isolating ones, such as ceramics and glasses—with a resolution in both the vertical and horizontal direction of below 1nm. Figure 1 shows a typical example of an AFM measurement in vacuum, depicting a cerium oxide surface in true atomic resolution. The inherent mechanical characteristics of AFM requires serial acquisition of data, which limits the speed for obtaining high-resolution images. Using commercially available microscopes, a typical image of 256 × 256 pixels resolution can be obtained in 1–10min. This is significantly slower than the time resolution required, for example, to resolve biological processes such as the movement of motor proteins in cells.²

Figure 1. Atomic force micrograph of a cerium oxide surface with true atomic resolution, imaged using a conventional atomic force microscope in noncontact mode.

Enhancing imaging speed demands improved scanner technology and control electronics. Specifically, several limiting factors with respect to the cantilever probe must be addressed to reduce resolution time. First, the measurement bandwidth of the local interaction between the probe tip and sample, as well as the velocity at which the tip moves, must be increased. Additionally, the tip must be able to follow the sample's topography at greater speed.³ Ideally, cantilever probes will function at resonance frequencies in the megahertz region with low force constants (approximately a few nano-Newtons per nanometer). Here, we report our progress in fabricating cantilever probes for high-speed imaging in AFM.⁴

We recognized that, to improve imaging speed, cantilever dimensions must be reduced to ∼0.1–1μm (thickness) × 3–20μm (length) × 2–6μm (width). Although such cantilevers have previously been reported, and applied successfully by different groups,^5–7 some issues with mass fabrication persist. For example, tiny cantilevers require an adequately shaped support chip to avoid contact between the chip corners and the sample under investigation. Additionally, precise control over the cantilever dimensions is crucial for integrating low-mass, high-aspect-ratio tips of appropriate length. Fine control of the tip is required to reduce squeezed-film damping between the sample and probe, thus enhancing the mechanical quality and sensitivity of the instrument. Furthermore, the curvature radius of the tip should be as small as possible.

We manufactured silicon support chips whose shape tolerated accidental lateral tilts greater than 5° without disrupting the requisite lengths of the cantilevers (10μm) and tips (2μm, see Figure 2). We controlled cantilever length with an accuracy of better than 2μm using either stepwise wet chemical etching or a cantilever-release technique, which we are currently developing. Cantilever length was determined by the rigid edge of the support chip, which also guaranteed its harmonic oscillation (important in noncontact applications). We integrated the tips—consisting of very hard, conducting, and almost completely sp³ hybridized (i.e., saturated) carbon—using electron beam-induced deposition (EBID). This approach enabled us to tightly control the height (up to 3μm), aspect ratios (better than 5:1), and radii (shorter than 10nm) of our tips. For improved performance, we compensated the mounting tilt of the probe within the AFM instrument using tips with axes tilted ∼8°.

Figure 2. (Left) Optical micrograph of our silicon support chip, with heavily slanted edges, for ultrashort cantilever probes. (Right) Scanning electron micrograph (SEM) of a silicon cantilever (top view), which protrudes from the small side of the support chip. The cantilever dimensions are 250nm (thickness) ×12μm (length) ×4μm (width). The bright point at the end of the cantilever is the tip, which was constructed using electron beam-induced deposition (EBID).

Having fabricated the cantilevers and tips using our silicon support chips, we explored a number of different applications for their use. By varying their composition and geometry, we adapted the cantilevers for operation in contact and noncontact modes, respectively. Cantilevers used in highly sensitive measurements operating in noncontact mode—in air or vacuum—must demonstrate high mechanical quality, harmonic oscillation, and have a force constant greater than 1.5N/m. These properties ensure that the cantilever does not abruptly snap near the sample surface. The quality factor of a dampened mechanical system—such as an AFM cantilever—is defined as the ratio of energy stored in the oscillating system to the energy dissipated per cycle.⁸ In ultrahigh vacuum, the cantilever quality factor is determined solely by intrinsic dampening mechanisms. Monocrystalline silicon is an ideal material for cantilever fabrication as it offers comparably low intrinsic dampening and near-zero volume loss.⁸ However, because of small variations in the manufacturing process, monocrystalline silicon limits cantilever thickness to ∼250nm. Despite this restriction, we successfully developed silicon probes with cantilevers thicker than 250nm, force constants greater than 1.5N/m, resonance frequencies of greater than 2MHz, and quality factors greater than 400 in air (see Figure 2). We found these cantilevers to be well-suited for noncontact measurements in air and vacuum.

Since many high-speed AFM applications focus on probing in contact mode or in liquids, we turned our attention to fabricating cantilevers from amorphous materials. These operations require cantilevers with force constants less than 0.5N/m at resonance frequencies greater than 1MHz in air. We developed two probe types comprising silicon-rich silicon nitride^9–11 and quartz-like thin films.¹² We used these amorphous materials because intrinsic dampening is negligible in contact mode or liquids, and clear harmonic behavior is less crucial for these operations. We used thin-film techniques to improve the thickness control (below 100nm) and yield. We also structured the layers independently from the support chip. Our thin films demonstrated high Young's moduli (a measure of stiffness), suitable for AFM cantilevers, without intrinsic stress gradients. Many amorphous thin films show plastic deformation under alternating loads and have an intrinsic stress gradient¹³ that causes cantilever bending. We overcame these obstacles by carefully controlling processing conditions, including source gases, gas ratio, temperature, and pressure. Figure 3 shows our AFM probes with cantilevers made of amorphous materials.

Figure 3. (Left) SEM of a cantilever—99nm (thickness) ×13μm (length) ×4μm (width)—composed of deposited amorphous silicon-rich silicon nitride thin film. At its end is a 2μm sharp tip, constructed using EBID. (Right) SEM of a cantilever—280nm (thickness) ×10μm (length) ×5μm (width)—with an EBID tip composed of a quartz-like material. The support chip of these probes is similar to that shown in Figure 2.

In summary, we developed three different cantilever probe types for high-speed AFM, which cover a wide range of possible applications. Based on these concepts, we will focus our future activities on the chemical and physical functionalization of small cantilevers, such as depositing different metal coatings for magnetic or electrical probe applications.