Piezoelectric materials have the unique ability to convert electrical signals into sound waves that can propagate along the material's surface or through its bulk. The inverse is also possible: Sound waves or mechanical vibrations on the surface or in the bulk of the material can be converted into electrical signals. For converting electrical signals into sound waves, piezoelectric crystals can be used to build structures that resonate within narrowly defined frequency ranges. These structures are ideally suited for highly selective filters in wireless communications systems.

The key is the ability to distinguish among a multitude of electrical signals and to transmit the appropriate signal to an individual user. Surface acoustic wave (SAW) devices are commonly used for this application. However, heating and cooling effects during operation of these devices affect their performance.

These materials can also be used to make delay lines, mixers, voltage-controlled oscillators (VCOs) and other basic building blocks of modern wireless networks. As for SAW filters, temperature stability is an increasingly important consideration as bandwidth requirements increase.

Currently, the design engineer must balance several compromises. For SAW applications, optimal thermal stability is typically achieved at the expense of coupling efficiency, and vice versa. Contemporary SAW filters, which are fabricated on a single piezoelectric substrate, exhibit these compromises with narrowed performance characteristics. One solution is to decouple stability from loss by using two piezoelectric materials — one for the surface device and a second to ensure thermal stability, and bond them together. This is the premise of temperature-compensated piezoelectric substrates. Early bonding techniques, however, were based on heating the materials, often leading to breakage and poor yields.

The development of ZiROC®, a room temperature, covalent bonding process, eliminates thermally induced issues and the need for critical design tradeoffs. An enabling technology to temperature-compensated piezoelectric devices, the ZiROC bonding process allows the engineer to select materials based on their specific characteristics and bond them together, thus optimizing the resulting substrate's thermal thermo-mechanical performance. Initial results have indicated a significant reduction in frequency drift, resulting from a factor of two improvements in temperature stability.

The piezoelectric effect was discovered by Pierre Curie in 1883. It demonstrates the mechanical distortion certain crystals undergo when a voltage is applied across this surface. The distortion is represented as tension and compression. When compressed or pulled, alternate charges form on opposite faces of the crystal, thus acting like a capacitor when a voltage is applied. A current, called piezoelectricity, is then generated between the faces. The converse is also true. When subjected to an external voltage, the crystal will expand or contract, launching a sound wave. The potential application to filters is described at right.

A simple antenna can be formed by placing two electrical contact pads on a piezoelectric material. If an electromagnetic field is applied, a sound wave, with a wavelength determined by the design of the contacts, will travel in the crystal. The velocity of the sound wave is a constant for a particular crystal, under a given set of conditions. With wavelength and velocity fixed, the wave equation,

λ × γ = υ

where λ is wavelength, γ, frequency, and υ, velocity of the wave, indicates that frequency will also be fixed. The powerful implication is, by controlling the dimension of the antenna and choosing a crystal that has the needed velocity, the antenna can be designed to select or to filter out a certain frequency. The antenna will, therefore, resonate at that particular frequency and will select that frequency component from a complex signal.

Frequency selection and control are critical considerations in wireless communications systems. When operating a cell phone or other wireless device, the environment is filled with electrical signals, and the device is trying to listen to one particular frequency. To get an appropriate signal-to-noise ratio where the system would be reliable, the transmitted power could be increased, but at a cost of shortened battery life. If, however, one of the filters is positioned between the antenna and the amplifier, all other frequencies that represent noise can be eliminated and the frequency of interest amplified. The filter effectively separates the signal of interest from the background noise, thus paving the way for very low-power communication systems with minimal trouble.

Cell phones must be able to operate in environments where temperatures could change by more than 100° F, in below freezing conditions, and under the hot summer sun. Cell phone manufacturers typically require components to operate from -35°C to 85°C. Small temperature changes within this range of even a few degrees can produce physical changes in piezoelectric materials and therefore, device characteristics. This tolerance to temperature is expressed in parts per million per degree Celsius. Cell phones using conventional, uncompensated, SAW devices have a thermal coefficient of frequency of ~45ppm/°C and may drift in frequency with even small temperature changes. For example, if the cell phone is designed to operate at a specific frequency and the temperature changes by 1°C, frequency divided into a million parts and multiplied by 45 of those parts determines the amount the frequency will shift. A decrease by a factor of two to 20ppm/°C would be a significant and welcomed improvement. This could be achieved with temperature-compensated piezoelectric materials, but understanding the impact of heat on these delicate crystals must first be addressed.

When the temperature of the filter changes, two things occur independently. The velocity of sound in the piezoelectric material changes, and the crystal gets physically larger. Because the antenna is made on the crystal, it, too, is stretched, changing dimension and, therefore, the wavelength of the traveling sound wave. This, in addition to heat-induced changes in velocity, changes the two previously fixed components in the wave equation, λ and υ. The third component, frequency, γ, must also change. When this happens, the frequency of the filter is drifting. This may force some filters out of specification required for systems, like cellular PCS, that operate in the gigahertz range with a system bandwidth of less than 100 kHz.

Currently, sharp antennas with sharp frequency cut-offs for both receiving and transmitting sound waves can be built. However, when the crystal heats, the center frequency of the filter can shift away from the center of the band established by the communications system's standards. To compensate for this, the engineer typically designs a wider filter to ensure the signal stays in the box, which, in turn, passes more noise.

As shown in Figure 1, a filter can be seen as a gate. An ideal filter operates over a narrow frequency range, has zero signal outside the band of interest and has an unattenuated or zero loss of signal. Unfortunately, when filters, including SAWs, are made wider to accommodate temperature drift, the gate gets correspondingly shorter, thus attenuating the signal. This attenuation is the insertion loss (the difference between the magnitude of the electrical signal in and the output signal). What was once a few percentage points loss in signal for a narrowly defined filter may become as high as 10% when designed wider. More power is required to boost the signal, thus affecting battery life. Additionally, wider frequency band filters limit the extent to which the spectrum can be partitioned, restricting support of more traffic.

The piezoelectric materials commonly used in SAW devices are single crystalline substrates of quartz, lithium tantalate (LiTaO3) or lithium niobate (LiNbO₃). The SAW device is formed on the surface of the substrate. Of the common piezoelectric materials currently in use, the expansion coefficient ranges from 12.6 ppm/°C in lithium tantalate to 10 ppm/°C for lithium Niobate. This number indicates a maximum expansion in a given crystallographic direction and also suggests the direction of highest coupling efficiency. Conversely, the ideal orientation of the antenna for thermal stability will produce the highest insertion loss. Relegated to the use of a single substrate, compromise is needed. Therefore, these crystals are sliced at an angle that lies intermediate to optimal stability and optimal couple, and tailored for a particular application. For example, if thermal stability is more critical, as in wireless communications, the “cut” direction favors stability, therefore, coupling efficiency as well as filter performance are sacrificed.

Thermal management of piezoelectric devices has traditionally fallen to the design engineer to compensate for dimensional changes in the antenna and to accommodate changes in the velocity of sound. However, the most direct and effective solution is to decouple the two parameters, allowing independent material selection based on optimal benefit. Early researchers in this field, however, ran into another temperature stumbling block as they attempted to bond a thin slice of one piezoelectric material, well suited for coupling efficiency, to a thermally stable substrate. The problem was in the heating process required to achieve bonding. Upon cooling, wafers would break, decimating manufacturing yields. Even more modest temperature excursions can induce substrate warping, which severely limits manufacturability. To achieve the necessary yield to make production-worthy, high-performance SAW devices requires a room temperature bonding process.

Once the optimal materials are chosen, the SAW device, a surface device, can be built on a thin (~<50µm) layer of piezoelectric crystal. This is simply achieved by bonding a full thickness piezoelectric wafer to the temperature-compensating layer and then thinning via back grind and chemical mechanical polishing (CMP). On heating, the thin skin wants to expand with temperature but it is now attached to material underneath, which expands minimally.

However, not one but two things are occurring. The dimensional change of the antenna is constrained; it expands less than it did before. And the amount it does expand, introduces a stress component. The speed of sound in the material is a function of stress (as well as temperature). The net result is that the filter changes frequency. This effect is seen when a carpenter saw is used to make music. By bending the saw and creating stress, the tone and thus the frequency, changes. This is what is occurring on a microscopic level in piezoelectric materials when a thin layer is bonded to a larger substrate. The stress induced in that structure, however, changes the velocity of sound to compensate for the temperature impact on the velocity of sound. These together enhance the thermal stability of the device.

The influence of temperature on frequency has been reduced and replaced with a stress component that influences a change in frequency but has a net positive affect. The thickness of the device layer controls the amount of stress introduced, which affects the velocity of sound, and therefore the frequency.

The ZiROC process yields a stress-free interface as initially bonded. Stress develops with temperature. Stress intruded during manufacturing will also influence operation. Ziptronix eliminates the built-in stress that results from high-temperature bonding and gives the engineer maximum freedom. Ziptronix bonded substrates can achieve a factor of two more thermal stability, allowing more aggressive filter designs for coupling efficiency and low loss filters, without sacrificing filter performance.

The SAW filter application is the logical starting point, given the sheer volume of SAW devices used in cellular handsets, global positioning systems (GPS), high-definition television and so on. To date, no commercial SAW devices based on temperature-compensated piezoelectric materials are available, though the demand for stable, sharply defined SAW filters is on the rise.

Other piezocrystal-based devices can also benefit from temperature-compensated substrates — Nyquist filters for microware digital radio, VCOs for first or second stage mixing in mobile transceivers, delay line-based path equalizers, and clock recovery filters for fiber-optics repeaters all require temperature compensation. These different classes of piezo devices offer elegant solutions to common problems in communications systems, simplifying circuit and system design and reducing cost.

Markets for networks that support high-bandwidth data services, such as 5 GHz 802.11a wireless local area networks (WLAN) standards and related follow-ons, are on track to provide a high volume opportunity for wireless devices. These high-bandwidth applications (well beyond current cellular frequencies) have tight frequency standards, a large range in signal strength, and are targeted for indoor use where multipath issues arise and crosstalk effects can dominate performance. Data-centric services will drive bandwidth requirements of WLANs to higher-frequency operation. These high-frequency applications will require increasing sophistication in the analog “front end.” Filters, frequency synthesis modules, clock synchronization and other devices for signal processing are fundamental tools of the system designer. Devices built on temperature-compensated piezomaterials offer low cost, high performance and ultra compact solutions for these system functions. This will be a major factor in the widespread adoption of high-bandwidth wireless networks.

Ziptronix bonding is a room-temperature process, requiring no application of pressure or voltage. No chemicals or adhesives are sandwiched between the surfaces, and no inert gases or vacuum is required. Its uniqueness lies in a surface preparation and chemical activation sequence. Called ZiROC, this proprietary process begins with a CMP to create the surface finish for bonding. The bonding layer may be any one of a number of common insulating layers, such as, silicon dioxide and silicon nitride, or in this case, piezoelectric materials.

Each surface is then treated with the proprietary ZiROC process, which activates formation of covalent bonds. The enhanced bonding effect induced by the chemical treatment persists for several hours, consistent with the requirements of high-volume, batch-mode production. Next, the prepared surfaces are brought together, forming a permanent bond. ZiROC-enabled covalent bonds are extremely strong and withstand downstream manufacturing process temperature excursions, as well as the thermal and mechanical stresses of wire bonding and flip-chip packaging techniques.

Room temperature bonding allows material combinations never before available to be achieved. The near ideality of the bond allows the properties of the constituent materials to determine the performance. The properties of the bond are simply not a significant factor. Thus material can be selected to combine strengths and cancel weaknesses. In short, a new class of materials can be “engineered” for a particular application.

The first of these applications is coming to market now. The list of applications is growing rapidly. System-level designs have long been used to mask various device instabilities such as temperature drift. For high-cost, niche-market applications, this strategy has been successful. We now see high-performance systems entering consumer markets where low cost is essential. Substrate engineering provides elegant solutions that simplify device design and system architectures — providing high performance at a cost consistent with high-volume consumer applications.

Robert Markunas, vice president of market development, Ziptronix Markunas spent the majority of his career with Research Triangle Institute International (RTI), in the role of director of the center for semiconductor research. Markunas holds a Bachelor of Science degree in Engineering from Massachusetts Institute of Technology, and participated in graduate studies in Electrical Engineering at MIT as well. Markunas has presented and published more than 90 works, and has 10 patents, with two others pending.

RSS    Save to Del.icio.us  Digg This