Chapter 10

Additional Topics

Introduction

In the previous chapters we talked about very many sensors and actuators of various types. The discussion focused on the principles of operation and on some of the applications of these sensors and actuators.

In this chapter we look at some additional aspects of sensors, aspects that could not have been discussed in conjunction with principles. First, we discuss a new class of sensors and actuators called MEMs (Micro-Electro-Mechanical). MEMs really relates to the method of production of sensors/actuators whereas the sensors and actuators themselves may be almost any of the sensors/actuators discussed previously. Then we tackle the issue of smart sensors that again includes many methods and many types of sensors but in general, smart sensors imply that additional electronics have been incorporated with the sensors. This may mean, for example, that a processor, an amplifier or some other type of circuitry has been incorporated with the sensor. Smart sensors are not always very “smart” in the sense of what they can do but are a step above regular active or passive sensors. Sometimes a smart sensor is a true necessity and evolved from the need to solve a problem. For example, a sensor may need to be very close to the processing circuit. It is then not a far stretch to bring the processor to the sensor and package them together. In other cases, the production method, especially with silicon based sensors, happens to be the same and therefore it is only natural that the two are combined. A third topic to be discussed is the issue of wireless. Although a sensor cannot be “wireless” the term wireless sensors is commonly used for sensors who communicate with the outside world through a wireless link. This approach is becoming more common and hence we will discuss not only issues associated with sensors but also wireless issues including frequencies, methods of modulation and issues of antennas and coverage including range. The final topic in this chapter is sensor arrays. We have already discussed sensor arrays, especially optical arrays. The meaning here is different. Whereas an optical sensor array was a sensor made of a number of individual sensors put together to perform a function (usually imaging), sensor arrays here will mean an array of sensor, spatially separated where each sensor performs its function and sends the data to a central location for processing. Thus, the main distinction is in the spatial distribution of the arrays.

1. Micro-Electro-Mechanical (MEMs) Sensors and Actuators

Micro-Electro-Mechanical sensors form a class of sensors that use two distinct properties; First, the sensors and actuators are produced using micromachining methods borrowed from semiconductor production methods. Second, the devices contain some sort of mechanical member such as a flexural beam, a diaphragm or, indeed, truly moving parts such as wheels and cogs. In a wider view, MEMs are viewed as any sensor/actuator which is micromachined, meaning that it is structured or sculpted out of a base material such as silicon by various means.

Here we will take a much narrower view, primarily for the purpose of narrowing down the subject and to avoid overlap with previously discussed material such as that in chapter 6, 7 and to a lesser extent in chapter 5 where various sensors such as semiconductor based pressure sensors as well as accelerometers and others, which are usually micromachined sensors were discussed. To do so we will concentrate here strictly on those sensors and actuators which employ moving members in the true sense including micromotors, micropositioners, grippers and the like (all are actuators) and other devices which employ these elements for sensing purposes.

However, before we do so, it is well worth discussing some topics in micromachining as well as basic techniques in semiconductor processing since these methods are at the base of all MEMs. A few of the methods used for construction of MEMs are:

1. Oxidation: a layer that can vary in thickness up to a few micrometers is created on the surface of semiconductors at high temperatures in preparation for processing and to create insulating layers as necessary. The process may be applied many times during a process.
2. Patterning: During various stages of production, various patterns need to be defined (such as conducting pads, areas of doping, shapes of transistors and the like). These are made using lithographic techniques in which a photoresist is placed on the silicon and exposed through a proper mask using UV sources. Following development, the pattern is created and then hardened by baking the remaining photoresist. Fig. 10.1 shows the basic steps involved in patterning.

Fig. 10.1. Basic steps in patterning.

3. Etching: Following patterning, sections may be removed using various types of etchants. For example, the pressure chambers in pressure sensors or the beam and mass in accelerometers are produced using this process. Various methods of etching and various etchants are used for specific purposes. Figure 10.2 shows some of the etching methods employed . Other methods, including plasma etching shown in Fig. 10.3 are used. In this method, rather than a chemical process, ions are used to bombard the exposed areas and remove the required material.

Fig. 10.2. Methods of etching.

Fig. 10.3. Methods of plasma etching

4. Doping: The production of various types of silicon (n, p, types) are also generated through the patterning masks as necessary. Again, there are various methods such as diffusion of dopants in the atmosphere around the wafer or ion implantation. Common materials are gases of phosphorus (n-type) or boron (p-type). Another method is ion implantation usinf the elements boron (p-type) and arsenic (n-type). Following Ion implantation, the material must be annealed to allow its atomic structure to relax into its final position.

5. Depositions: In the production process it is often necessary to deposit various layers of materials. Thin films of silicon and other materials, including metals, are deposited at various thicknesses. There are many methods of deposition but the most common methods are based on chemical vapor deposition whereby the wafer is placed in an atmosphere containing the vapor of the material to be deposited. Metal deposition includes aluminum and gold as well as others. Figure 10.4 shows two common systems for metal deposition – the epitaxial system and the low pressure chemical vapor deposition (LPCVD)

a.b.

Fig. 10.4. a. The epitaxial and b. LPCVD deposition systems

6. Bonding: A range of bonding techniques at various stages of the production. In some cases bonding simply means bonding the silicon wafer to a substrate or package but in other cases bonding is used to seal chambers (such as in gauge pressure sensors or absolute pressure sensors). Bonding can be done by gluing with bonding agents, fusing of silicon to silicon, bonding to glass and many others at low and high temperatures.

The processes described above are common integrated circuit techniques and for the basis of semiconductors production. However, for MEMs productions additional techniques are needed. These are temed micromachining techniques. Many of these are methods that allow construction of structural members of the MEM. Some of the most common micromachining methods are as follows:

1. Bulk micromachining: The wafer is deep etched from the back of the structure and usually takes advantage of special etchants for different layes. One approach is to use etchants whose rate of etching depends on the orientation of the silicon crystal. Also, various methods of stopping the etching at the desired depth are used to control the structure. These methods can vary from simply timing the etching process to inserting specific materials to stop the process at the required depth. This method allows creation of deep chambers as well as structural members such as diaphragms, beams, etc. Deep etching of a membrane is shown in Figure 10.5 which also indicates the different rates of etching on different crystal cuts.

Fig. 10.5. Deep etching of a membrane

2. Surface micromachining: Here the process takes place on the surface of the wafer. Typically, additional layers of metals or, more often, polysilicon are used in conjunction with other materials to fabricate mechanical components. For example, a sacrificial layer may be deposited on the surface on which the polysilicon is deposited followed by removal of the sacrificial layer to expose the mechanical member (Fig. 10.6).

Fig. 10.6. Basic surface micromachining process (cross-sectional (i) and lateral (ii) views)

There are many variations of these methods and combination of methods that allow fabrication of very fine structure of various depths and thicknesses but as a rule, these are based on the methods described above or combinations thereof.

In addition to fabrication methods borrowed from silicon processing, there are so-called non-silicon technologies. Some of these such as LIGA and EPON are capable of producing very slender, high aspect ratio structures such as cogs and wheels at a scale that is typically larger than common semiconductor scales (a few mm). Figure 10.7a shows a LIGA sacrificial process and Figure 10.7b shows a product fabricated using the EPON process.

a. b.

Fig. 10.7. a. The LIGA sacrificial process. B. Product made using the EPON process.

The devices produces using any or all of these methods are often integrated with semiconductor circuits to produce sensors/actuators. An example may be a pressure sensor produced using bulk micromachining followed (or preceded) by production of semiconductor strain gauges and/or amplifiers or other processing elements.

MEMs Actuators

As mentioned above. MEMs for the purpose of this chapter are those that contain physically moving members beyong the simple flexing of a beam or a membrane. In terms of actuators, this means that we are dealing with micro-motors of various types as well as other devices such as micro-grippers, escape mechanisms and the like.

In the very early days of development of MEMs, a variety of very interesting devices were built, including rotating motors based on electrostatic attraction/repulsion. These “motors” while perfectly valid as mechanisms were used to demonstrate the technology. They can only produce minuscule forces and torques (although they can rotate very fast) and therefore are not likely to replace any existing actuation technologies but they may find limited applications in miniature robots. Nevertheless, MEMs have found use in some very important actuation applications. Some of these are as follows:

Micromotors

A variety of micromotors have been produced using various fabrication techniques, some of them having diameters of a few hundred micrometers. These motors are electrostatic in nature, that is, the forces are attraction forces between stator plates and rotor plates as shown in Figure 10.8. By properly modulating the voltages on the plates, induced charges on the rotors affect the forces and the motion. Typical construction is in two polysilicon layers and produced in a few steps over a substrate as shown in Figure 10.9.

Fig. 10.8. An electrostatic micromotor.

Fig. 10.9. Typical construction of micromotors

The same methods are used for other types of structure such as the micro-locking system shown in Figure 10.10. Here use is being made of wheels produced as for rotating motors but with limited motion and with latching mechanisms which can be actuated using electrostatic forces or by other means (magnetic, bi-metal, heating, etc.).

Fig. 10.10. A micro-locking system.

Resonator and comb drives

Another successful method of mechanically driving mechanism is the comb structure shown in Figure 10.11. Here, each side of the structure is a comb like capacitor. The comb structure increases the capacitance and the attraction/repulsion of the central movable structure is responsible for the motion. The central structure is anchored. The structure can be moved slowly or it can be made to resonate mechanically. This idea has been applied to a number of devices including gyroscopes and to drive rotors through escape mechanisms.

Fig. 10.11. A comb resonator. Note the anchors that provide restoring forces.

Micro-pumps

Another interesting application is shown in Figure 10.12. Here a microchannel is built in silicon together with two valves. A method of expanding the diaphragm must be provided. This can be electrostatic or piezoelectric, or, as shown, it may be thermal using a heater to heat a gas or fluid in a closed chamber which, upon expansion generates a pressure differential. The fluid is expelled through the sealing outlet valve. In the cooling process, the fluid is sucked into the pumping chamber. Clearly this is not a high power device but it may be used for medicine delivery systems and for low volume metering applications.

Fig. 10.12. A micropump actuated by heat. Note the two valves.

Deformable mirrors

In light modulation for displays and projectors, it is often sufficient to deform a mirror or rotate it a minute angle to affect significant changes in reflection of light. In these applications mechanical force is minimal and therefore it is a good candidate for implementation in MEMs. An example is shown in Figure 10.13. Here the forces are electrostatic. The deformation can be as in Figure 10.14 in which an aluminum coating forms the reflection surface. Another method, adopted in TIs projection system is shown in Figure 10.15. This particular device has been implemented into a full 3D projection system in which each color is projected separately (Figure 10.16).

Fig. 10.13. Electrostatic forces used to deform the surface of a mirror. The mirror is usually an aluminum deposition layer.

Fig. 10.14. Another method of deforming a mirror.

Figure 10.15. Tilting mirror activated by electrostatic forces and used in a projection system as an array of mirrors.

Fig. 10.16. TI’s projection system using deformable mirrors. Each chip is an array of mirrors.

Microtips and manipulators.

The methods used for production of MEMs have yielded some interesting and useful devices. One of them is a microtip, produced by lithographic techniques. One of these, shown in Figure 10.17. These microtips have been used for sensors and in scanning electron microscopes for manipulation on very small scales, including manipulation of individual atoms because of their very sharp tip.

Fig. 10.17. A very fine microtip used for micro-manipulation.

Another is the microgripper shown in Figure 10.18. These devices are envisioned for use in microassembly and microrobots, including possible applications in remote work with hazardous substances and in medicine.

Fig. 10.18. An electrostatic microgripper. Note the scale.

Microconveyor belt. Here the idea is that although any MEM is very small and cannot develop any significant force, an array of these devices can generate large forces and hence produce useful work. An example is the microconveyor belt shown in Figure 10.19. The operation is simple: the upper part and the lower part of the belt are separated by movable segments which can tilt, affecting rotation, and hence linear translation in either direction as shown in Figure 10.20.

Fig. 10.19. Photograph of a microconveyor belt

Fig. 10.20. Operation of the microconveyor belt showing the lifting and rotation motion of the actuators against the fixed (lower) and movable (upper) layers.

2. Smart sensors/actuators

A smart sensor or actuator is any sensor/actuator in which some level of “intelligence” has been introduced. This simply means that in addition to the normal function of the device, circuitry has been added to take on such functions of local data processing and, sometimes, even decision making. That also means that additional power for the electronic circuits must be available for the purpose. For example, a microprocessor may be added to the sensor/actuator to analyze the data, perhaps to digitize the output from the sensor, to compensate for unwanted stimuli and, of course, to communicate to the main processor to which the sensor is ultimately connected. The level of intelligence, or how “smart” the sensor is may vary from the trivial to the truly complex. At the lower end, this may include such simple circuits as current and voltage limiters, active filters and compensation circuits. In the high end of the spectrum, all functions of processing, including digitization, data transmission (wire or wireless), data logging and any conceivable function may be included on board making the sensor a free standing sensing system. In actuators, protection circuits such as thermal protection, overvoltage and overcurrent protection as well as motion and limit functions may be added. Other options are counters, alarms, data loggers and many others. Integration of electronics with sensors, particularly silicon based sensors is always possible but it is based on the commercial viability of the sensor. Sometimes, this integration makes perfect sense, especially when they are used for mass market application (the automobile industry, toys, for example). In other cases it is better to leave the sensor as a general purpose sensor and allow the designer to integrate it in the design as necessary.

Figure 10.21 shows a basic smart sensor and its connection to a controller through a 4-20 mA loop (to be discussed in the following chapter). To be noted are the microprocessor and, in this case, a separate Analog to Digital Converter (ADC – also to be discussed in the following chapter). While the exact configuration and needs will vary from one sensor to the other, this configuration is representative of the type of devices one can expect to find in a smart sensor.

Fig. 10.21. A smart sensor fed by a 4-20 mA loop. The loop also transfers the data to the controller. The data given are for specific components from a manufacturer.

3. Wireless Sensors and Issues Associated with Wireless Sensors

Sensors, of course are not, by themselves, wireless. The term wireless sensor refers more to the link with the sensor than with the sensor itself. In most cases, a communication link is made available to send data out and, sometimes, to receive data. This of course means that the sensor must, by necessity be a smart sensors since most sensors do not produce data that can be sent directly over a wireless link. For example, a thermocouple produces a dc signal (or a slowly varying signal). This signal must first be digitized and then used to modulate a carrier before it can be transmitter. On the other end, at the processor, the opposite process must take place. Some sensors are truly digital and produce, for example, a signal whose frequency is proportional to the stimulus. These are usually easier to interface with wireless systems. All this is fairly obvious and represents essentially replacement of the physical link with a wireless link. However, it also means that remote sensing in the true sense can now take place. The sensor (or, for that matter an actuator or both), may be far from the processor, perhaps on a different continent, in space or on a different planet. The communication link itself may be a short range link or may use microwave communication links, wireless communication systems or satellites, depending on the need.

Many sensing systems make use of short range wireless communication, often in a dedicated link. These typically use one of the ISM (Industrial, Scientific and Medical) bands allocated by the FCC (or by the ISO (International Standards Organization) in Europe and other countries. These frequencies are allocated for general use and unregulated (except for frequencies and power allowed). These are used by very many applications such as remote control (garage door openers, keyless entry in vehicles and buildings), for hobby and for data transfer. Operation in these bands is strictly enforces as to frequency, bandwidth and transmitted power allowed. This also means that the range is short – typically less than 100 m, often much less. Nevertheless this range is often sufficient for remote sensing within a building, around the factory floor and the like. In many cases, the necessary range is so small that even an induction link is sufficient but, nevertheless, this is a wireless link.

The ISM and SRD bands

In the United States, the Federal Communications Commission (FCC) is responsible for allocation of frequencies for general use by industry and the public. In Europe and many other countries, the European Telecommunications Standards Institute, (ETSI), the European Radiocommunication Office (ERO) and the CISPR Comission of the International Telecommunication Organization (ITO) regulate the use of frequencies. The allocations in the US (and Canada) and those in Europe and other countries are not the same but, at least in part overlap.

The ISM (Industrial, Scientific and Medical) band was originally allocated to be used in industry for such things as microwave ovens, dielectric welders and the like as well as in medical applications including microwave treatment of tumors. These frequencies are shown in Table 1. The low frequencies are commonly used in industrial microwave heating, welding and cooking but also in RFID tagging and short range communication.. Other frequencies, such as the 2.45 GHz band are used in consumer products such as microwave ovens. The 915 MHz band is universally used for communication and control as well as for RFID applications and many more. Some of the allocated frequencies have no current use but have been incorporated for future use.

Table 1. ISM allocations, uses and allowable power.

Frequency	Some typical Applications	Power/field strength
124 - 135 kHz	low frequency, inductive coupling, RFIDs, tire pressure sensing	72 dBmA/m
6.765 - 6.795 MHz	inductive coupling, RFIDs	42 dBµA/m
7.400 - 8.800 MHz	Article surveillance	9 dBµA/m
13.553 - 13.567 MHz	inductive coupling, contactless smartcards, smartlabels, item management, dielectric welding, short range communication	42 dBµA/m
26.957 - 27.283 MHz	Industrial microwave ovens, dielectric welding	42 dBµA/m
40.660 - 40.700 MHz	Industrial microwave ovens, dielectric welding	42 dBµA/m
33.050-434.79	Remote entry, wireless controll	10 – 100 mW
2.400 - 2.483 GHz	Remote control, vehicle identification, microwave ovens, LANs, Bluetooth, WLAN, ZigBee, cordless telephones	4 W - spread spectrum, USA/Canada only, 500 mW, Europe
5.725 - 5.875 GHz	Wireless video cameras for security, wireless communication, control, WiMAX, future use	4 W USA/Canada, 500 mW Europe
24.000 – 24.25 GHz	To be used in the future	4 W USA/Canada, 500 mW Europe

Notes:

1. The last three bands are divided into channels each 0.5 MHz in bandwidth.

2. The power/field values are for communication applications. In microwave ovens the power is much higher since the system is enclosed.

The SRD (Short Range Devices) have been allocated for what is often called “unregulated” use. Unregulated is a misnomer because they are actually very tightly regulated but they are available for use without special licenses by any product, as long as they conform to the provisions of the regulation in terms of frequency, bandwidth, power, and often, duty cycle. The SRD frequencies are shown in Table 2. Of these, the only truly internationally accepted frequency is the 433 MHz. This is almost universally used for short range control (keyless entry system, garage door openers). In the US, other bands have been used in the past for this purpose (290 MHz, 310, 315 and 418 MHz) but the tendency is to conform more and more with international allocations. The higher bands (860 MHz to 928 MHz) are still separate and there does not seem to be any convergence to common bands.

It should be noted that each frequency band comes with its own constraints. For example, the higher frequencies (860 MHz and higher) usually incorporate bands. One can operate within one band or hop between them but cannot use a bandwidth that spans two bands. In the 433 MHz range, the bandwidth is fixed, there are no bands (i.e. a single band), power is constrained to 10 mW or up to 100 mW under special license and the duty cycle is larger than 10%. That is, one is allowed to transmit for up to 1 sec at a time with a minimum of 10 seconds between transmissions.

Table 2. SRD allocations, uses and allowable power.

Frequency	Some applications		Power
433.050-434.79 MHz		See Table 1	10 mW
863.0-870.0 MHz		Various, including wireless audio, alarms, RFIDs, cell phones	5 – 500 mW depending on band
902.5 - 928 MHz		Same as above	4 W - spread spectrum, USA/Canada only

Sensors and actuators often operate in the ranges mentioned above either as separate systems or in conjunction with other systems. For example, RFID systems are being used extensively for identification and tagging of items ranging from consumer products to pets. In that capacity they identify the item and contribute to efficient distribution and tracking of products. However, they are also very important in sensing. For example, many cars currently employ a key identification system whereby only the key that has been properly programmed can be used. The system employs an RFID in the key and a transceiver in the car dashboard or steering columns senses the presence and identifies the key. What is unique here is the very short range (usually below 1 m) and the wireless communication at low frequencies (typically 13 MHz). Other RFIDs employ a true sensor. For example, Microchip produces an RFID with a sensor input to allow temperature sensors to monitor perishable products such as meat on their way to market. As said above, there is nothing particularly ingenious in incorporating sensors in wireless systems once the special needs of signal processing imposed by the wireless system have been taken care of.

4. Sensor networks

Our discussion so far concentrated on sensors and actuators operating independently or, in some cases, multiple sensors being connected to affect a larger output (thermocouples, for example). However, there are systems which are much more complex than these and require a number of sensors (sometimes a large number) in a distributed configuration. These may be used to sense a distributed stimulus, sometimes over a very large area. The output of these sensors can then be used to take decisions or to operate appropriate actuators to affect a required function. For example, a series of sensors may be used to monitor the watershed of a river and, perhaps automatically open spill gates on dams to avoid flooding and to prevent damage. Other applications are in sensing hazardous materials, traffic sensing and control, wireless security networks in public areas and many others.

In a system of this type, the network is fairly simple. Each sensor performs its sensing function and transmits the data to a processor by an appropriate communication link, including wireless. A system of this type is used in the US and Canada to monitor lightning strikes, nationally. The system employs detectors which detect the radio waves transmitted by the lightning strike. By using directive antennas the direction of the lightning strike relative to the sensor is identified. The distance to the lightning strike is found from the amplitude of the signal received at the sensor. By incorporating a system of sensors throughout the US and sending the signals back to a central location, a map of lightning strikes is obtained and used in weather prediction (see www.strikestarus.com). Figure 10.22 shows the sensor system and some data it can provide. Sensor networks for detecting and predicting thunderstorms are also in place. The sensors in these systems are obviously smart sensors in the true sense.

Fig. 10.22. A map of the US and Canada showing locations of lightning sensors and various data that can be obtained from this sensor network

However, sensor networks can be much more complicated than the ones described so far. In particular, in conjunction with low power wireless applications, there is a whole methodology of incorporating sensors and methods of communication so that each sensor can only communicate to a few of its neighbors, perhaps only the nearest neighbors. In large systems, it then becomes necessary to process data locally on each sensor and only communicate the absolute necessary data. The reason for this is that normally, processing of data is fast and predictable whereas communication is slow and prone to interference and loss of data.