Parallel scanning probe arrays: their applications
aMcCormick School of Engineering and Applied Science, Northwestern University, Evanston, IL, USA
Since the invention of the scanning tunneling microscope (STM)1 and the atomic force microscope (AFM)2, the field of scanning probe microscopy (SPM) instruments has grown steadily and has had a profound influence in materials research, chemistry, biology, nanotechnology, and electronics[3] and [4]. Today, scanning probe instruments are used for metrology, characterization5, detection6, manipulation7, patterning[8] and [9], and material modification. A wide range of scanning probe applications are available, taking advantage of various modes of tip–substrate interactions, including force, optics[10] and [11], electrochemistry12, electromagnetics, electrostatics, thermal and mass transfer[13] and [14], and vibration[15] and [16].
The scanning probe instrument family includes various surface characterization tools that measure surface force interactions; these ‘force microscopy’ tools include AFMs2, magnetic force microscopes[17] and [18], electrostatic force microscopes19, lateral force microscopy[20] and [21], and so on. The AFM has been used to characterize surfaces of inorganic materials, organic materials, and biological entities.
Scanning probes can also be used to produce high-resolution spatial mapping of topography, hardness, temperature, emitted or reflected light, charge distribution22, and vibration magnitude16. SPM probes have also been widely used in surface modification, in either additive mode23, subtractive mode, electrochemical reaction mode12, or thermal phase change mode24. Since scanning probes are essentially nanoeffectors connected to high-precision mechanical movement controllers, they have been used as manipulators7.
Today, advancements in nanoscience and nanotechnology are being pursued in a multidisciplinary fashion and on a global scale25. An SPM has broad technical appeal because it embodies a number of powerful and yet turnkey features: sharp end effectors, a computer-programmable mechanical motion stage, a calibrated precision force actuator, and a high-sensitivity motion detector26.
The scanning probe instruments play a key role in facilitating the top-down as well as bottom-up agendas of nanomanufacturing. SPMs will be increasingly used to address needs beyond laboratory research. Examples of future needs include large area metrology, high throughput characterization and detection6, high density data storage27, and large area nanolithography.
Traditional scanning probes use a singular tip/cantilever entity. This poses a limit on the throughput of imaging and manipulation tasks. Linear scan rates are typically of the order of 1–10 μm/s. At 10 μm/s, it would take or 2.4 h to cover a distance of 10 cm, the diameter of a typical wafer. In order to increase the throughput and area coverage, it is important to use parallel arrayed probes, preferably high-density and large-area arrays. The parallel scanning probe array, which is a chip or substrate containing at least two scanning probes engaged in a serial or parallel operation, is crucial for satisfying future research and industry needs.
This paper will illustrate fundamentals of parallel scanning probe microscopy – including design and fabrication of probes, integration of functions, and operations. Further, it will use a few cases to exemplify how parallel scanning probes are designed, made, and used.
Basics of scanning probes
The probes in an SPM system are one of the crucial elements affecting performance, functionality, and speed. An SPM generally consists of a cantilever with a tip located at the distal end (Fig. 1). The cantilever is in turn connected to a handle for handling and mounting in instruments.
Today, SPMs are often made of microelectromechanical systems (MEMS) technology, which is uniquely capable of miniaturization (giving rise to low force constant and high resonant frequency), precision (giving rise to repeatability and uniformity), mass production on a wafer scale (giving rise to low cost), and electromechanical functional integration28.
The flexural displacement of cantilevers in force-microscopy measurements is often sensed optically with an external light source and detector. (Alternatively, the torsional displacement of the beam along its longitudinal axis may be used29.) A laser beam is reflected off the cantilever to a light detector; movement of the reflected spot indicates the extent of flexural bending. The displacement of the cantilever can be measured by using other methods, such as light interference30 or surface-stress sensing elements (including piezoresistors, piezoelectric sensors, and stress-sensitive transistors[31] and [32]). Piezoresistors may be realized using doped silicon, whereas piezoelectric sensors are made by depositing and patterning piezoelectric materials including lead zirconate titanate (PZT) and ZnO.
The flexural bending mode of the cantilever is most commonly encountered. When a force F acts on the end of a cantilever with a length of l, a cross-section of w × t, and a material Young's modulus of E, it produces a tip displacement d and a surface stress s. The magnitude of d and s are 
and 
, respectively28. The equivalent spring constant of the beam is 
. The first-order natural frequency of the cantilever is 3.57 
, with ρ being the density of the cantilever material.
Design compromises are necessary in many cases. Certain applications demand probes with low force constant (for high sensitivity) and high mechanical resonant frequency (for high speed). Increasing the length of the cantilever, for example, would tend to reduce the force constant, increase the surface-induced stress (thus increasing sensitivity if integrated sensing is used), and reducing the resonant frequency. Reducing the thickness would make the cantilever more compliant, reduce the resonant frequency, and increase the force sensitivity. One must carefully select the design and materials to obtain the desired performance characteristics.
Basics of SPM probe fabrication
The art of realizing an individual cantilever, without the tip attachment, is relatively well established in the MEMS area. Cantilevers may be made of a variety of materials, including silicon nitride (Si3N4), single-crystalline Si, polymer, and metal28.
SPM probes are more complex compared with bare cantilevers. The need to incorporate tips, especially sharp or high-aspect ratio tips, increases the degree of difficulty of processing beyond that for plain cantilevers. Namely, the process must yield sharp tips first, preserve the tip sharpness throughout the process and accomplish high process yield, while allowing a broad choice of tip and cantilever materials. There are other areas of complication. For example, certain probes may require integration of sensors and actuators.
Tips made of Si and Si3N4 (by chemical vapor deposition) are most common today. Some applications may demand tip materials that are unconventional, i.e. not compatible with traditional MEMS and microelectronics processes. Other demonstrated materials for tips include metal33, elastomer34, and diamond35.
There are a number of important technical issues that may seem easy to solve. The cantilever needs to have sufficient optical reflectivity (if optical sensing is used) and controlled intrinsic bending. Intrinsic bending is due to stress mismatch between layers at different thicknesses. It is undesirable for a number of reasons: (1) If the bending is not controllable or repeatable, the implementation may require tedious optical alignment. (2) Intrinsic bending causes the force–displacement characteristics of the probe to change, altering the effective force constant and enhancing the transverse displacement.
For many applications, it is desirable for the tip to face in the opposite direction to the handle, such that the handle does not impede the operation of the SPM. Another challenge is the packaging of such devices – the release of individual elements and drying of finished MEMS devices, which contain delicate cantilevers, are often nontrivial tasks.
Processes can be categorized according to a number of criteria. Processes may be discrete or monolithic. A discrete process involves precision assembly of various discrete elements (e.g. tips or cantilevers) into a joint device. A monolithic process is defined by the fact that all elements are made from a contiguous material, without needing delicate assembly or attachment. Monolithic processes are preferred for their uniformity and efficiency. Discrete processes may offer certain flexibility but are difficult to achieve due to the small sizes of the elements.
Monolithic fabrication processes can be categorized in many ways. The processes fall into two major categories according to the method of tip fabrication. In the first category, tips are made by etching. In the second category, tips are made by molding.
Tip etching can be accomplished by both dry and wet chemistry. This typically results in very sharp tips; however, the uniformity and repeatability is questionable. Tip formation by molding can result in greater uniformity of geometry and sharpness. However, tip molding is accomplished by etching away the substrate, which is wasteful and time consuming, or by removal of a sacrificial space layer, which increases the radius of curvature of the tips.
Tips may be sharpened by a variety of methods, including oxidation followed by subsequent oxide removal, by ion beam etching, or by growth of high-aspect-ratio nanostructures. For example, C nanotubes may be grown at the end of tips using chemical vapor deposition36.
The available processes can also be categorized by the material of the cantilevers (insulator or conductor).
Commonly used routes for the fabrication of nonsensorized scanning probe cantilevers are summarized in Fig. 2. General method (a) starts with a single-crystal silicon wafer (often <100> oriented) (Fig. 2a.1). Cavities for molding tips are made by anisotropic etching of Si (Fig. 2a.2); common etchants include KOH and EDP[28] and [33]. The materials for tips and cantilevers are then deposited and patterned (Fig. 2a.3), followed by the attachment of a handle piece (Fig. 2a.4). Finally, the mold material is removed (Fig. 2a.5). General method (b) also starts with a single-crystal Si piece (Fig. 2b.1). A convex tip is formed by etching (e.g. anisotropic chemical etching, plasma etching) (Fig. 2b.3). The back side of the wafer is patterned in preparation for a bulk etch (Fig. 2b.4), after which a cantilever with controlled thickness is left standing (Fig. 2b.5). Method (c) is similar to method (b) in the first two steps. The material is coated with a thin film (Fig. 2c.3), which serves as the tip and cantilever. Bulk etching is completed by chemical etching from the front side of the wafer (Fig. 2c.5).
Parallel probes: design, materials, and fabrication
MEMS technology is uniquely suited for fabricating parallel scanning probes. If MEMS and photolithography are used, making an array of n probes does not take n times the efforts of making a single probe. In addition, MEMS can increase the degree of uniformity of device geometries. However, there are unique challenges for parallel arrays, in addition to those encountered for individual probes. (1) The maximum density of tips is dictated by the minimal distance between probes. The decision of minimal distance is affected by cantilever geometries, footprint for wiring, and cross-talks among neighboring probes. (2) The uniformity of tip and cantilever dimensions, of mechanical properties, and of tip height is of the utmost importance and must be carefully controlled through design, processing, and materials selection. The degree of intrinsic bending is related to the dimensions of the cantilever, the stress level of various layers, and the thickness (Fig. 3). (3) The processing yield must be extremely high. If the chance of process failure for an individual probe is x, with x being a number between 0 and 1, the chance of process failure for an array of n probes is 1-(1-x)n.
