Microengineering Aerospace Systems
Henry Helvajian, editor
Chapter 1: Introduction to MEMS
M. Mehregany and S. Roy
1.2 Fabrication Technologies
The three characteristic features of MEMS fabrication technologies are miniaturization, multiplicity, and microelectronics. Miniaturization is clearly an important part of MEMS, since materials and components that are relatively small and light enable compact and quick-response devices. Multiplicity refers to the batch fabrication inherent in semiconductor processing. Consequently, it is feasible to fabricate thousands or millions of components as easily and concurrently as one component, thereby ensuring low unit component cost. Furthermore, multiplicity provides flexibility in solving mechanical problems by enabling the possibility of a distributed approach through use of (coupled) arrays of micromechanical devices. Finally, microelectronics provides the intelligence to MEMS and allows the monolithic merger of sensors, actuators, and logic to build closed-loop feedback components and systems.
Clearly, the successful miniaturization and multiplicity of traditional electronics systems would not have been possible without IC fabrication technology. It is therefore natural that the IC fabrication technology, or microfabrication, has so far been the primary enabling technology for the development of MEMS. Microfabrication provides a powerful tool for batch processing and miniaturization of mechanical systems into a dimensional domain not accessible by conventional (machining) techniques. Furthermore, microfabrication provides an opportunity for integration of mechanical systems with electronics to develop high-performance, closed-loop-controlled MEMS. Integrated fabrication techniques, which are made possible by IC fabrication technology, eliminate the need for discrete component assembly, which is not practical for the fabrication of MEMS. Hence, dimensional control, including component size and intercomponent clearance, is limited only by the processing technology.
Even though miniaturization of mechanical systems is directly compared to that of electronics, two important points should be noted. First, not all mechanical systems will benefit from miniaturization. More likely, microfabricated sensors and actuators that enable performance improvement will be integrated into conventional macroscopic mechanical systems. Miniaturization and the application of microtransducers for monitoring and control is justified when the performance-to-cost ratio is improved. Second, current IC-based fabrication technologies are inherently planar, not allowing full flexibility for three-dimensional (3D) design. A mature technology for micromechanical systems will require complementary fabrication techniques that provide3D design capabilities.
1.2.1 IC Fabrication
Any discussion of MEMS first requires a basic understanding of IC fabrication technology. The major steps in this technology include film growth, doping, lithography, etching, dicing, and packaging (see Fig. 1.2). Devices are usually fabricated on Si substrates, which are grown in boules, sliced into wafers, and polished. Thin films are grown on these substrates and are used to build active components, passive components, and interconnections between circuits. These films include: (1) epitaxial Si, (2) SiO2, (3) silicon nitride (Si3N4), (4) polycrystalline Si (polysilicon), and (5) metal films. To modify electrical or mechanical properties, films are doped with impurities by thermal diffusion or ion implantation. Lithography is used to transfer a pattern from a mask to a film via a photosensitive chemical called a photoresist. The film is then selectively etched away, leaving the desired pattern in the film. This cycle is repeated until fabrication is complete. The wafers are then probed for yield, diced into chips, and packaged as final devices. Because there is a market for high-quality, inexpensive Si wafers (namely, microelectronics), most MEMS fabrication facilities focus on thin-film growth, doping, lithography, and etching processes.
|
Fig. 1.2. Major processing steps in integrated circuit fabrication. |
1.2.1.1 Film Growth
The growth of SiO2 by the thermal oxidation of Si is the fundamental film growth process. In IC fabrication, oxidation is used for passivating the Si surface, masking diffusion and ion implantation layers, growing dielectric films, and providing an interface between Si and other materials. In MEMS, SiO2 films are also used as etch masks and sacrificial layers, which will be discussed later. Although Si exposed to air at room temperature will grow a native oxide (about 20 Å thick), thicker oxide films (0.5–1.5 µm) can be grown at elevated temperatures in a mixture of hydrogen (H2) and oxygen (O2) gases. The rate of oxide growth is dependent on the growth temperature, the oxygen partial pressure, and the crystal orientation of the Si substrate. However, for a fixed temperature, oxide thickness increases with time in parabolic fashion.
To deposit SiO2 films on substrates other than Si, a process known as chemical vapor deposition (CVD) is used. In this process, the chemical components of the film are supplied to the reactor as a mixture of gases. The substrate is heated to a temperature that induces a pyrochemical reaction and film formation. Such depositions are performed at atmospheric pressure (AP) or low pressure (LP). In most instances, the process is ideal for batch coating. Growth rates are much higher in APCVD systems, while films are deposited with excellent thickness uniformity in LPCVD systems. CVD is also used to deposit thick (>1.5 µm) oxide films or when the substrate cannot be simply oxidized thermally. In addition to SiO2, metals, polysilicon, Si3N4, and many other films can be deposited by CVD.
Epitaxial growth is a special class of CVD. Epitaxy is defined as growth of a single crystal film upon a single crystal substrate. If the composition of the film is the same as that of the substrate, the process is called homoepitaxy. However, if the film composition differs from that of the substrate, the process is called heteroepitaxy. Many compound semiconductors, such as gallium arsenide (GaAs) and silicon carbide (SiC), can be grown heteroepitaxially on Si, while doped Si layers are homoepitaxial.
Many films are not thermally stable at temperatures commonly used in conventional CVD. To reduce deposition temperatures so that existing films will not be adversely affected, CVD pyrochemical reactions are combined with a radio-frequency (RF) plasma in a process known as plasma-enhanced CVD (PECVD). PECVD results in films with good step coverage and low pinhole density. However, PECVD films generally suffer from significant hydrogen incorporation and lower mass density, degrading the electrical and mechanical properties of a material.
Metal films can be deposited by vacuum evaporation, sputtering, CVD, and plating, and are most commonly used for interconnections, ohmic contacts, and rectifying metal-semiconductor contacts. Vacuum evaporation is used to deposit single-element conductors, resistors, and dielectrics. Alloys can also be deposited by this method, but the process is complicated by the widely varying evaporation rates of different metals. Resistive and electron beam heating are the two most common heat sources. Compound materials and refractive metals can be deposited by sputtering a cathode target with positive ions from an inert gas discharge. Introduction of noninert gases into the ambient during sputtering is called reactive sputtering and is used to deposit compound films such as titanium nickel (TiNi).
1.2.1.2 Doping
In many instances, it is desirable to modulate the properties of a device layer by introducing a low and controllable level of impurity atoms into the layer. This process is called doping and is accomplished by either thermal diffusion or ion implantation. Thermal diffusion is performed by heating the wafers in a high-temperature furnace and passing a dopant-containing carrier gas across the wafer. The diffusion process occurs in two stages: predeposition and drive-in. During predeposition, dopant atoms are transported from the source onto the wafer surface and are diffused into the near-surface region. The sources can be gaseous (e.g., diborane [B2H6]) or solid (e.g., boron nitride [BN]), depending on the dopant. During drive-in, the temperature is increased, and the dopant diffuses into the wafer to the desired depth and concentration. Ion implantation introduces dopants below the wafer surface by bombardment with an energetic beam of dopant ions. Because the energy loss of these ions in Si is well known, precise control of the dose and depth of dopants is possible. The crystal lattice is damaged during this process, but the damage can often be reduced by subjecting the wafer to a high-temperature, postimplant anneal.
1.2.1.3 Lithography
Lithography is the technique by which the pattern on a mask is transferred to a film or substrate surface via a radiation-sensitive material. The radiation may be optical, x-ray, electron beam, or ion beam. For optical exposure, the radiation-sensitive material is more commonly called "photoresist", and the process is called "photolithography." Photolithography consists of two key steps: (1) pattern generation and (2) pattern transfer. Pattern generation begins with mask design and layout using computer-aided design (CAD) software, from which a mask set is manufactured. A typical mask consists of a glass plate coated with a patterned chromium (Cr) film. Pattern transfer involves: (1) dehydration and priming of the surface, (2) photoresist coating of the wafer, (3) "soft bake" of the photoresist, (4) exposure of the photoresist through the mask, (5) chemical development of the photoresist, (6) wafer inspection, and (7) postdevelopment bake or "hard bake." After hard bake, the mask pattern has been completely transferred to the photoresist .
1.2.1.4 Etching
Following hard bake, the desired pattern is transferred from the photoresist to the underlying film or wafer by a process known as etching. Etching is defined as the selective removal of unwanted regions of a film or substrate and is used to delineate patterns, remove surface damage, clean the surface, and fabricate 3D structures. Semiconductors, metals, and insulators can all be etched with the appropriate etchant. The two main categories of etching are wet-chemical and dry-etching. As the name implies, wet-chemical etching involves the use of liquid reactants to etch the desired material. However, tighter governmental regulations on safety and waste, coupled with the trend toward smaller device features, have led to an increasing emphasis on dry etching. There are various types of dry-etch processes, ranging from physical sputtering and ion-beam milling to chemical-plasma etching. Reactive ion etching, the most common dry-etch technique, uses a plasma of reactant gases to etch the wafer, and thus is performed at low pressure in a vacuum chamber. Well-characterized wet-chemical and dry-etch recipes for most semiconductor processing materials can be found in the literature and will not be detailed here.
|
Fig. 1.3. Bulk micromachined features realized by anisotropic etching of silicon. (a) Bottom plan view of etched wafer with cavities, diaphragms, and holes; (b) top plan view of an anisotropically etched wafer showing the fabrication of a cantilever beam using etch stop layer; (c) cross section, AA', showing the hole, diaphragm, and cavity of (a); and (d) cross section, BB', showing the cantilever beam of (b). |
In order to fabricate structures, etching is used in conjunction with photolithographically patterned etch masks. The effectiveness of an etchant depends on its selectivity, that is, its ability to effectively etch the exposed layer without significantly etching the masking layer. Since most etch masks are not completely impervious to etchants, mask thicknesses depend on the selectivity of the etchant and the total etch time. Suitable etch-mask materials for many dry- and wet-chemical etchants include SiO2, Si3N4, and hard-baked photoresist.
1.2.2 Bulk Micromachining and Wafer Bonding
Bulk micromachining was developed between 1970 and 1980, as an extension of IC technology, for fabrication of 3D structures.14 Bulk micromachining of Si uses wet- and dry-etching techniques in conjunction with etch masks and etch stops to sculpt micromechanical devices from the Si substrate. There are two key capabilities that make bulk micromachining of Si a viable technology. First, anisotropic etchants of Si, such as ethylene-diamine and pyrocatechol (EDP), potassium hydroxide (KOH), and hydrazine (N2H4), are available that preferentially etch single crystal Si along given crystal planes. Second, etch masks and etch-stop techniques are available that can be used in conjunction with Si anisotropic etchants to selectively prevent regions of Si from being etched. As a result, it is possible to fabricate microstructures in an Si substrate by appropriately combining etch masks and etch-stop patterns with anisotropic etchants.
|
Fig. 1.4. Bulk micromachined cantilever fabricated by p+ etch stop and anisotropic etching. |
Good etch masks for typical anisotropic etchants are provided by SiO2, Si3N4, and some metallic thin films such as Cr and Au. These etch masks protect areas of Si from etching and define the initial geometry of the region to be etched. Alternatively, etch stops can be used to define the microstructure thickness. Two techniques for etch stopping have been widely used in conjunction with anisotropic etching in Si. One technique that uses heavily boron (B)-doped Si, called "p+ etch stop," is effective in practically stopping the etch. Another technique, called "pn junction," stops the etch when one side of a reverse-biased junction diode is etched away.
Anisotropic wet etchants of Si, such as KOH, are able to etch Si <100> and <110> crystal planes significantly faster than the <111> crystal planes. In a <100> Si substrate, etching proceeds along the (100) planes but is practically stopped along the <111> planes. Since the <111> crystal planes make a 54.7-deg angle with the <100> planes, slanted walls result, as shown in Fig. 1.3. Because of the slanted <111> planes, the size of the etch-mask opening determines the final size of the etched hole or cavity. If the etch mask openings are rectangular and the sides are aligned with the [110] direction, practically no undercutting of the etch-mask feature takes place. However, significant undercutting below the mask may occur in convex corners (corners with angles greater than 180 deg), where the etch masks are misaligned with the [110] direction, or where there are curved edges in the etch-mask openings. Under these circumstances, the undercutting continues until it is limited by the <111> planes. Undercutting can be used to fabricate suspended microstructures. Figure 1.4 shows a bulk micromachined Si cantilever fabricated by undercutting the beam's convex corners-defined by an etch stop-from the front side of the wafer.
|
Fig. 1.5. Complex shapes patterned using deep reactive ion etching (DRIE). |
A drawback of wet anisotropic etching is that the microstructure geometry is defined by the internal crystalline structure of the substrate. Consequently, fabricating multiple, interconnected micromechanical structures of free-form geometry is often difficult or impossible. Two additional processing technologies have extended the range of traditional bulk micromachining technology: deep anisotropic dry etching and wafer bonding. Deep anisotropic dry etching of Si can be achieved using reactive gas plasmas, which will etch exposed Si vertically. Recent improvements in this technology allow the patterning and etching of high-aspect-ratio (e.g., 20:1), anisotropic, randomly shaped features into a single crystal Si wafer, with only photoresist as an etch mask.15 As shown in Fig. 1.5, etch depths of a few hundred microns into an Si wafer are possible while maintaining smooth, vertical sidewall profiles. The other technology, wafer bonding, permits an Si substrate to be attached to another substrate, typically Si or glass. Electrostatic (or anodic) bonding of Si to glass substrates is performed under application of pressure and high voltage (400–1000 V), while Si fusion bonding (SFB) is the bonding of two Si wafers at high temperatures (near 1000°C), in an O2 or N2 ambient. By combining anisotropic etching and wafer bonding techniques, bulk micromachining technology can be used to construct 3D complex microstructures such as microvalves and micropumps. Figure 1.6 presents a microvalve that is fabricated by anisotropic etching and bonding of four Si wafers. In addition to dry etching and wafer bonding, the capabilities of bulk micromachining are further enhanced by laser processing techniques (Chapter 5) applied to microstructures up to 1-mm thick with 20:1 aspect ratios.
|
Fig. 1.6. Schematic cross section of a microvalve fabricated by bulk micromachining and wafer bonding. The TiNi shape memory film is thermally actuated to open and close the microvalve (H. Kahn, Case Western Reserve University). |
1.2.3 Surface Micromachining
Surface micromachining relies on encasing specific structural parts of a device in layers of a sacrificial material during the fabrication process. The sacrificial material is then dissolved in a chemical etchant that does not attack the structural parts. In surface micromachining, the substrate wafer is used primarily as a mechanical support on which multiple, alternating layers of structural and sacrificial material are deposited and patterned to realize micromechanical structures. Surface micromachining enables the fabrication of complex, multicomponent, integrated micromechanical structures that would be impossible with traditional bulk micromachining.
A typical surface micromachining process, shown in Fig. 1.7, begins with the deposition of a sacrificial layer, which is then patterned to create openings to the underlying substrate. Next, the structural layer is deposited and patterned into the desired geometry. Finally, the structural components are released by removal of the underlying and surrounding sacrificial material. The structural components are attached to the underlying substrate at the anchor regions.
|
Fig. 1.7. Cross-sectional schematic demonstration of surface micromachining. (a) Sacrificial layer deposition, (b) definition of the anchor and bushing regions, (c) structural layer patterning, and (d) free-standing microstructure after release. |
Surface micromachining is a versatile technology for three key reasons. First, the patterning of the structural and sacrificial layers is typically accomplished by etching processes that are insensitive to the crystalline structure of the films, thereby providing flexibility for planar free-form designs. Second, surface micromachining enables integrated multilevel structures using multiple layers of structural and sacrificial material. Third, there is no express restriction on the structural-sacrificial material system, as long as the compatibility between the structural and sacrificial materials is maintained. Therefore, different application-specific structural layers can be used in conjunction with suitable sacrificial layers.
Polysilicon surface micromachining using polysilicon as the structural material and SiO2 as the sacrificial material has been the most widely used surface micromachining technique. When electrical isolation of the substrate and/or the structural components is required, Si3N4 is used as an insulator. In this process, hydrofluoric acid (HF) is used to dissolve the sacrificial oxide during release. Figure 1.8 presents a surface-micromachined shear-stress microsensor fabricated using a single structural layer of polysilicon. Another commercially used surface micromachining technique utilizes aluminum (Al) and photoresist as the structural and sacrificial layers, respectively. In this case, the release of the structural Al layer is accomplished by removing the sacrificial photoresist using a plasma etch. A number of other material systems have also been investigated as structural/sacrificial layers for surface micromachining: Al/polyimide, Si3N4/polysilicon, and Si3N4/SiO2. The maximum thickness of structural layers in traditional surface micromachining is limited to 10 µm or less because of residual stresses in films. Excessive residual stress can lead to mechanical failure during fabrication. Furthermore, there are process limitations due to slow film deposition rates in traditional methods such as CVD, sputtering, and evaporation. Faster deposition rates can be realized for films that can be grown using pulsed laser deposition (PLD) or plating techniques.
|
Fig. 1.8. SEM of a shear-stress microsensor fabricated by surface micromachining using a single structural layer of polysilicon. |
1.2.4 Micromolding
Micromolding refers to fabrication of microstructures using molds to define the deposition of the structural layer. After the structural layer deposition, the final microfabricated components are realized when the mold is dissolved in a chemical etchant that does not attack the structural material. Micromolding is an additive process, in that the structural material is deposited only in those areas constituting the microdevice structure. In contrast, bulk and surface micromachining are examples of subtractive micromachining processes, which feature blanket deposition of the structural material followed by etching to realize the final device geometry.
A widely known micromolding process is Lithographie, Galvanoformung, und Abformung (LIGA). This German acronym means lithography, electroplating, and molding. The process can be used for the manufacture of high-aspect-ratio, 3D microstructures in a wide variety of materials (e.g., metals, polymers, ceramics, and glasses).16,17 As shown in Fig. 1.9, high-intensity, low-divergence, hard x rays are used as the exposure source for the lithography. These x rays are usually produced by a synchrotron radiation source. Polymethylmethacrylate (PMMA) is used as the x-ray resist. Thicknesses of several hundreds of microns and aspect ratios of more than 100 have been achieved. A characteristic x-ray wavelength of 0.2 nm allows the transfer of a pattern from a high-contrast x-ray mask into a resist layer with a thickness of up to 1000 µm so that a resist relief may be generated with an extremely high depth-to-width ratio. The openings in the patterned resist can be preferentially plated with metal, yielding a highly accurate complementary replica of the original resist pattern. The mold is then dissolved away to leave behind plated structures with sidewalls that are vertical and smooth. It is also possible to use the plated metal structures as an injection mold for plastic resins. After curing, the metallic mold is removed, leaving behind microreplicas of the original pattern. By combining LIGA with the use of a sacrificial layer, it is also possible to realize free-standing micromechanical components.18
|
Fig. 1.9. Outline of the micromolding process using LIGA technology. (a) Photoresist patterning, (b) electroplating of metal, (c) resist removal, and (d) molded plastic components. |
A chief drawback of the LIGA process is the need for a short-wavelength collimated x-ray source like a synchrotron. Consequently, LIGA-like processes using conventional exposure sources are being developed. Photoresists with high transparency and high viscosity can be used to achieve a single-coating mold thickness in the range of 15 to 500 µm.19–21 Thicker photoresist layers may be realized by multiple coatings. In such photoresist layers, standard ultraviolet (UV) photolithography is used to achieve mold features with aspect ratios as high as 11:1.
Photosensitive polyimides are also used for fabricating plating molds.22 The photolithography process is similar to conventional photolithography, except that polyimide works as a negative resist. In this process, about 10-µm-wide lines can be delineated in several tens of microns-thick resist. A maximum aspect ratio of 8:1 can be achieved, but depends on the geometry of the mask layout. Polyimide is a very stable material and does not have to be cured to act as a plating mold, but it is also limited in terms of the thickness and the aspect ratio.
All methods stated above make use of lithography techniques to make a mold, but dry etching of polyimides to form high-aspect-ratio molds has also been reported.23 In these methods, some modifications of traditional reactive ion etching (RIE) systems are necessary to achieve high-aspect ratios. For example, dry etching of fluorinated polyimides with a titanium (Ti) mask has been used for deep etching with high-aspect ratios, excellent mask selectivity, and smooth sidewalls.23,24
Using micromolding processes, it is possible to realize high-aspect-ratio metallic microstructures, which are especially attractive for certain applications, including reflective surfaces for optical components, low resistivity contacts for relays, magnetic metals for electromagnetic actuators/sensors, and microfabricated coils. Additionally, the larger thickness of high-aspect-ratio structures provides for greater stiffness perpendicular to the substrate, as well as for increased force/torque in electrostatic actuators. Plated nickel (Ni), copper (Cu), or alloys that contain at least one of these metals are the structural metals commonly used; Cr, SiO2, polyimide, photoresist, and Ti have been often used as the sacrificial material.
Next: 1.3 MEMS Components
