Nico de Rooij
Institute of Microtechnology
University of Neuchtel
Engineers have been building mechanical elements and devices with dimensions in the hundreds of micrometers or MEMS (microelectromechanical systems) since the early 1980s. They wanted to exploit certain properties inherent in small parts: smaller mass means fewer mechanical restraints and more stability; small surface-to-volume ratios lead to faster thermal changes; and building parts in parallel, a common feature of microfabrication techniques, lowers production costs. And thanks to research into building better semiconductors, MEMS pioneers have the tools they need to observe and build micrometer-sized devices widely used in several industries. That is, they turned microscience into microtechnology.
Nanoscience, the study of devices measured in nanometers (10 -9 m) is undergoing a similar evolution. Most scientists and engineers involved in nanoscience are devising methods of measuring, manipulating, and observing nanoscale events and features rather than developing commercial or industrial-applications. Almost all of these tools have their origins in microtechnology.
FROM MICROTECHNOLOGY TO NANOSCIENCE
Nanotools need not have nanometer-scale dimensions, but they do require micro to millimetersize components to act as alligator clips or interfaces between the nano and macro worlds. Consequently, nanotools fall squarely in the MEMS domain and can be made using microfabrication. Here are some of the newer nanotools:
Dynamic-force sensor. An atomic-force microscope (AFM) measures the deflection of a small, cantilevered beam or sensor, which measures height at specific spots on a sample. In most cases, the beam deflects in response to stimuli and that deflection is detected by bouncing a laser off the beam and noting changes in the reflected laser. In a dynamic AFM, the cantilever is vibrated at its resonance frequency. The frequency characteristics change in response to stimuli, letting researchers determine topology and other sample properties.
A new dynamic AFM sensor uses a U-shaped, silicon cantilever connected to a quartz-tuning fork (QTF). Each of the cantilevers legs is attached to one of the tuning forks tines, and the other end of the cantilever has a probe tip. When the fork vibrates at its first resonance, the cantilever vibrates at the same frequency but perpendicularly to the motion of the two tines. If the tip moves close to the sample, the cantilever changes its vibrating frequency and the fork, mechanically coupled to the cantilever, becomes detuned. This lets engineers examining nanoscale objects and events specify the operating frequency by choosing the right tuning fork and specify a force constant by properly designing the cantilever. They can choose a high tuning-fork frequency to maximize the number of contacts between sensor tip and sample, which also increases the microscopes scan rate. Higher frequencies also mean more sensitive detection. They can also design cantilevers with lower spring constants. This reduces the force the probe exerts on the sample, which prevents scratches, and causes less wear on the tip.
Multifunctional AFM-probes. Investigating proteins in plant and animal cells with AFM requires cantilevered-spring probes that have soft spring constants (less than 0.1 N/m) and work in buffer solutions. The probes need low spring constants to move easily without damaging the sample or distorting the measurements. And studying electronic transport phenomena across cell membranes requires an electrically conductive probe tip that can measure small interactions or electrical phenomenon between the tip and the surface. To avoid leakage currents through the buffer, we encapsulate a titanium adhesion layer coated with platinum, the electrical contact, and tip in silicon nitride (SixNy) and SiO2, respectively. Silicon nitride is an insulating compound that can be laid down using chemical-vapor deposition. If the entire tip and cantilever were conductive, the current observed would be too high, as would the signal-to-noise ratio. With only the tip sensitive to current, the signal-to-noise ratio drops and current is only measured in the confined area of interest. The exposed metallic tip apex is about 90 nm high with a diameter at the base of about 90 nm.
Cantilever SNOM-probe. Scanning near-field optical microscopy (SNOM) sidesteps the diffraction limit on normal microscopes by using subwavelength light emitted from a 100-nm or smaller aperture in a probe held close to the sample. Resolution is determined by aperture size and the distance between probe and sample. Unfortunately, SNOM always relied on optical-fiber tips. These tips cannot be mass produced and, therefore, will always be expensive. To bring down the price and widen the availability of SNOM, researchers have developed a cantilever-based SNOM probe that can be mass produced. It is based on a Si cantilever with a silicon-oxide tip. This tip has a 60-nm-thick coating of aluminum and transmits light even without an aperture. Researchers are trying to discover how this is done. Despite the fact theres no physical opening on the tip, the light goes through it, giving researchers an optical resolutions power of about 32 nm. The resolution also improves because the probe is a sharp tip, much sharper than if it had one or more holes. With such tips, researchers will be able to image single molecules tagged with fluorescent labels.
Micropipette tips. Hollow tips are needed for dispensing or sampling liquids. Researchers are depositing a silicon-nitride layer in a silicon mold, then etching away the silicon.
From these examples of tools for nanoscience, one can see that most interactions with the nanoworld are mostly limited to surface phenomenon. We are still far from manipulating molecules in a bulk material to change material properties on a larger scale.
Unlike conventional mechanical engineering and metalcutting technology, microtechnology relies on chemical and photographic processes to shape materials. Photolithography, for example, starts with a pattern or outline of the device being built. The pattern, consisting of open areas where material is to be removed, and opaque areas for the actual part, is imprinted on the raw material using photosensitive chemicals. Later, material in open areas is removed. Several iterations of this produce a usable part. The technique lets designers build devices with features on the order of 1
m m. Hundreds of these patterns can be placed and etched from the same block or disk of raw material, letting devices be mass produced.
Silicon is the material of choice for many microfabrication techniques due to its mechanical, optical, and electrical characteristics. And silicon can be grown in large monocrystals and selectively etched relative to certain crystal planes in a wet potassium-hydroxide solution (KOH). This particular technique works well to construct cantilevered beams with small, sharp depositions or tips, a common component in microdevices.
Other techniques use reactive plasmas to anisotropically etch paths independent of crystal orientation. To increase the aspect ratio (height-to-diameter or height-to-width ratio) of such paths, some machines rapidly switch between different gases that alternately passivate sidewalls then etch away at the bottom surface of the path. Critical parameters of bulk etching are etching rates, compatibility of the raw material with the lithography process, and sidewall roughness.
Engineers also use thermal evaporation, sputtering, and chemical-vapor deposition to lay down thin films that are later selectively and chemically etched away or left alone to build silicon actuators or sensors. By using silicon with different doping levels, this method can create piezoresistive sensors that detect mechanical stresses and thermal actuators. Thin-film deposition and selective etching can also create electrical paths connecting sensors and passivation layers that protect and insulate them. Characteristics such as deposition rates, film stress, and surface roughness have to be controlled for best results. Thus, small mechanical elements become more elaborate devices or MEMS (microelectromechanical systems) with built-in sensors and actuators that use thermal, electrostatic, and magnetic forces.
Although silicon is the most widely used material for building MEMS devices, others can be processed as well. Borosilicate glass or quartz wafers, for example, are used in applications involving photosensitive detection where transparency is necessary. Borosilicates dielectric properties also let it handle high electric fields. And its coefficient of thermal expansion is close to that of silicon, so both materials can be bonded together to create hermetically sealed interfaces. This technique is often used to package MEMS.