WO2012085334A1

WO2012085334A1 - Method for manufacturing an ultrasonic sensor

Info

Publication number: WO2012085334A1
Application number: PCT/FI2011/051082
Authority: WO
Inventors: Markku Ylilammi; Aarne Oja; Jaakko Saarilahti
Original assignee: Teknologian Tutkimuskeskus Vtt
Priority date: 2010-12-21
Filing date: 2011-12-07
Publication date: 2012-06-28
Also published as: FI20106360A; FI20106360A0

Abstract

The invention relates to an ultrasound sensor(1) and a manufacturing method, the sensor (1) comprising an ultrasound transmitter (102, 103, 104), a cavity (4) arranged in connection with this, which is in resonance mode at the ultrasound frequency used,and the sensor (1) comprises a passive sensor element (3, 3'), the distance of which from the ultrasound transmitter (2) is selected in such a way that the resonance condition is met, located at the opposite end of the cavity (4) to the ultrasound transmitter (102, 103, 104). In accordance with the invention the ultrasound transmitter (102, 103, 04) is a piezo-element.

Description

Method for manufacturing an ultrasonic sensor

The present invention relates to a method for manufacturing an ultrasonic sensor according to the preamble to Claim 1.

The invention relates also to a sensor structure. 1. General In micromechanical pressure sensors, microphones, etc., the distance between two diaphragms is measured in a so-called capacitive manner. This can be done by measuring the voltage induced by movement, or by measuring the capacitance at a high frequency. Because the diaphragms are very close to each other, this leads to several problems: the diaphragms adhering to each other, movements caused by the thermal expansion of the base and a component, etc. In addition to mechanical problems, present constructions suffer from the fact that the impedance between the two diaphragms is high, and that all the electrical leaks cause problems.

In mobile telephones, a changeover is presently being made to micromechanical microphones. MEMS pressure sensors have already become generally used, for example, in watches. In terms of the MEMS construction, they are very similar. They differ only in packaging technique, selection of materials, electronic solutions, etc., but are nevertheless based on the same basic principle: measurement of the distance between two diaphragms located close (1 um - 3 μιη) to each other. These are referred to as first-generation MEMS microphones and pressure sensors. The present new invention eliminates several weaknesses of the first-generation sensors and makes possible a stable MEMS sensor for measuring several variables, which can be easily attached to a circuit card. Present gas sensors are either chemical or optical. Chemical sensors have poor stability, and optical sensors are expensive for mass applications.

A solution is known from NO patent 323259, in which an ultrasound transceiver, made using MEMS technology, is located in a tuned cavity. With the aid of the transceiver, the speed of sound in a gas is measured, and the properties of the gaseous substance are determined based on the speed of sound. The device has the drawback of a complicated construction, which demands two active elements to be formed, together with circuitry. The complexity increases manufacturing costs.

A gas-sensor application, in which a piezo -ultrasound transmitter is used, is known from the publication "Modelling of an electroacoustic gas sensor", F. Granstedt, M. Folke, M. Ekstrom , B. Hok and Y. Backlund, Sensors and Actuators B 104 (2005) 308-311.

A gas-sensor application, in which a piezo -ultrasound transmitter is used, is also known from the publication "Gas sensor with electroacoustically coupled resonator", F. Granstedt, M. Folke, B. Hok and Y. Backlund, Sensors and Actuators B 78 (2001) 161-165.

Because a prior art piezo -ultrasound transmitter is based on the radiation of a fixed oscillator, its connection to the gas is weak, in other words, the changes taking place in the gas or oscillation chamber appear faintly in the input impedance of the ultrasound transmitter, or correspondingly as a function of the changes in the power that it draws. In the articles referred to, only the speed of gas in a cavity is measured, no reference being made to the measurement of the location of a passive sensor diaphragm, based on a single active component.

EP- application 08856883 describes structures and different uses for an ultrasonic sensor. The manufacturing processes according to this application require

complicated steps for creating electric contacts for the ultrasound source.

The present invention is intended to eliminate drawbacks of the prior art and present a novel method for manufacturing an ultrasonic sensor.

The invention is based on forming the ultrasound transmitter as a piezo-element. More specifically, the method according to the invention is characterized by what is stated in the characterizing portion of Claim 1. The sensor in accordance with the invention is characterized by what is stated in claim 8.

Considerable advantages are gained with the aid of the invention. Compared to capacitive constructions, the electronics can be effectively shielded, because electrical contacts are not required to the diaphragm or surface acting as the sensor element.

Thus, the sensor can be brought into direct contact with the measurement object. In some preferred embodiments of the invention, even mechanical contact with the surface being measured is possible. Compared to the above described piezo-crystal ultrasound sources described above as the prior art, a significant improvement in sensitivity is achieved, because the light diaphragm ultrasound transmitter according to the invention connects about 10 times better to the medium being measured, this being apparent directly in the sensitivity of the measurement. At the same time, a significant cost advantage is achieved through the fabrication technique, because the sensor structure can be integrated, for example, in the same SOI structure.

Compared to the MEMS construction described, a significant advantage is achieved in the simplification of the device. Only one active element is required, with the aid of which the most diverse measuring elements can be created.

The manufacturing process is simpler that in conventional non-piezo solutions.

In the following, the invention is examined with the aid of examples and with reference to the accompanying drawings. Figures la and lb and lc show cross-sectional side views of alternative basic constructions of the sensor.

Figures 2a and 2b show schematically an electrical equivalent circuit of the sensor.

invention.

Figure 3 shows a cross-sectional side view of the construction of an acceleration sensor..

Figure 4 shows a cross-sectional side view of details of the manufacturing process of the sensor.

Figure 5 shows a cross-sectional side view of an alternative construction of the sensor. Figure 6 shows schematically the electronics of the sensor.

Figure 7 shows a cross-sectional side view of another alternative construction of the sensor.

Figure 8 show one graph for modelling the sensor in accordance with the invention. Figure 9 shows a cross sectional view of a simplified structure in accordance with the invention for analysing the invention,

Figure 10 shows a cross sectional view of a detail of simplified structure in accordance with the invention for analysing the invention,

Figure 11 shows a graph of relation of piezo width and corresponding amplitude. Figure 12 shows a graph of relation of piezo thickness and corresponding amplitude. Figure 13 shows a graph of relation of membrane thickness and corresponding amplitude. Figure 14 shows a graph of relation of membrane width and corresponding amplitude. Figure 15 shows a graph of relation of driving voltage and corresponding amplitude.

Figure 16 shows as a cross sectional view of a first process step for manufacturing a sensor in accordance with the invention.

Figure 17 shows as a cross sectional view of a second process step for manufacturing a sensor in accordance with the invention. Figure 18 shows as a cross sectional view of a third process step for manufacturing a sensor in accordance with the invention.

Figure 19 shows as a cross sectional view of a fourth process step for manufacturing a sensor in accordance with the invention.

Figure 20 shows as a cross sectional view of a fifth process step for manufacturing a sensor in accordance with the invention.

Figure 21 shows as a cross sectional view of a sixth process step for manufacturing a sensor in accordance with the invention.

Figure 22 shows as a top view a electrode and piezo structure of the sensor in accordance with figure 21. By measuring the impedance of the ultrasound through the input impedance or input power of the transmitter, the impedance will be observed to be strongly dependent on the deviation of the cavity above it relative to the aforementioned length. If operation takes place in the series resonance of the ultrasound, small changes in the impedance of the transmitter will be directly proportional to the height of the cavity and the width of the resonance will be proportional to the air damping in the cavity. The geometry of the sensor is arranged in such a way that the variable being measured affects the length of the cavity. Figures la and lb show the schematic construction of the ultrasound sensor 1. In Figure la there is a construction, in which the ultrasound transmitter 2 is formed of a single cavity and in lb there is a situation, in which sound is created using several oscillations transmitters 2 in parallel. The diaphragm 9 of the transmitter 2 creates a resonating ultrasound in a cavity 4 above it. The ultrasound transmitter 2 creates both electrical series and parallel resonance. The impedance of the series resonance is typically 1 - 10 kI2 and the impedance of the parallel resonance is 10 - 100-times greater, always according to the quality factor of the resonator. It can be assumed that the flexible upper diaphragm 3 will reflect most of the sound back and only a small part will radiate out behind the diaphragm 3. In practice, this means that the impedance of the series resonance diminishes. If the diaphragm radiates power out, the quality factor of the cavity diminishes and the resolution decreases. Figure lc shows a way, in which an additional cavity 6, the size of half the wavelength, and the cover 7 of which is preferably thick and stiff to effectively reflect the wave of ultrasound back into the cavity, is added above the cavity.

Figure 2a gives the electrical equivalent circuit of the ultrasound transmitter 2. The coil 23 depicts the mass of the diaphragm 9 and the capacitance 24 its spring constant and capacitance 20 the electrical capacitance. The resistance 22 represents the diaphragm's losses and the resistance 21 the radiation losses. The values of the components are also dependent on the voltage over the ultrasound sensor. If the sound cavity is, for example, λ / 2 _m length, it will amplify the resonance of the ultrasound. The sound wave arrives at the ultrasound diaphragm in phase, which increases the movement of the diaphragm. Figure 2b shows the electrical equivalent circuit with the cavity 4, when the cavity is λ / 2 _m length, i.e. the cavity forms a series resonance. It should be noted that the cavity could also be a quarter of the wave, or a multiple of it in length. The impedance of the cavity 4 replaces the radiation resistance 21 in Figure 2a. Thus, the impedance of the cavity is formed from the effective coil 25 of the resonator 4, the effective capacitor 26, and the losses 27 of the resonator. We note that the system forms a single new series resonance. We can consider the resonances as degenerating to form a single resonance. If the effective capacitance 26 formed by the cavity 4 is now smaller than the effective capacitance 24 of the ultrasound transmitter 2, and if the inductance 25 of the cavity 4 is greater than the inductance 23 of the ultrasound sensor, the impedance of the sensor 1 (= transmitter 2 + cavity 4) will be determined almost entirely by the cavity 4. In practice, the situation is not as ideal as this. In addition, we can assume that the internal losses of the ultrasound transmitter 2 are smaller than the losses created by the cavity 4. This can be achieved, if we use a micromechanical transmitter 2. It is not possible to create the situation in question using a piezo-active transmitter in accordance with the prior art. Optimization requires the mass of the oscillating diaphragm 9 of our ultrasound transmitter 2 to be minimized and the spring constant (tension or stiffness) to be optimized in such a way that the desired resonance frequency is achieved. In addition, the surface area of the diaphragm 9 should be dimensioned to optimize the connection with the gas. The optimization depends on whether the diaphragm oscillates in piston mode or in flexible mode. In addition, the shape of the transmitter 2 is significant. The diaphragm 9 can also be formed from many small individual ultrasound diaphragms 9. This situation is shown in Figure lb. In general, the radiation optimum will be achieved when the width of the diaphragm 9 is about in the order of the wavelength of the ultrasound frequency used. I.e., the natural dimensions of the cavity 9 are typically: width λ and height λ 12. The width can also be greater, if the transmitter is made from several separate elements. The location of the upper diaphragm 9 changes the phase of the impedance or, in other words, its imaginary part. The losses of the cavity 4 change the real part of the impedance. In other words, by using a phase-sensitive detector to measure the value of the imaginary part, we can determine the deviation of the diaphragm 9. I.e., as a point of departure the sensor 1 is a sensor that measures location or change in location, but which can be used to measure several different variables. Hereinafter we will refer to this sensor generally as the UtraSensor.

2. Applications

As a point of departure, the sensor 1 measures the location of the diaphragm, or the change in its location. This means that it can be used in all applications, in which precise measurement of the location of the surface is essential. In other words, the UtraSensor can be used to make a microphone, pressure sensor, acceleration sensor, or magnetometer. In addition, the device can be used to measure the speed of sound and the damping of a gas, i.e. it can be used as a gas sensor. If the cavity has the length of a quarter of a wavelength, and if it is filled with a gas that is heavier than air, it will form an impedance converter between the MEMS ultrasound transmitter and the air. In the following, the special features of various sensors are reviewed.

2.1 Location sensor

The UtraSensor is, as such, a sensor that measures location, velocity, or acceleration. Acoustic emissions are typically measured by connecting a piezo-active structure to the surface of a metal, to detect acoustic waves. In this application, the UtraSensor can be used directly to replace the prior art piezo sensors. The UtraSensor has the advantages of a smaller size and price and the possibility of integrating several receivers on the same silicon substrate. Of course, the location sensor can be used as a normal ultrasound sensor and transmitter, for instance, in ultrasound imaging devices.

2.2 Impedance-matched broadband ultrasound transmitter and receiver.

If we make the height of the cavity equal to one quarter of the wave, and we fill it with a suitable gas mixture (e.g. argon and carbon dioxide, nitrogen, ...), we can adjust the gas density in such a way that the specific impedance of the

micromechanical transmitter completely matches air. This takes place in practice, if the density of the gas in the cavity is about ten times greater than that of air. In practice, the gases, which can be used, and the density of which is significantly greater than that of air, are Argon and C0₂. Benzene can also be used. This solution would make possible, for example, a radar based on ultrasound that is better than existing ones (e.g., parking radar for cars), a velocity meter for gases, etc. In addition, because the electronics can be taken far from the subject being measured, it permits measurement in difficult conditions (e.g., a temperature of 300 degrees). Nowadays, the radiation efficiency from a MEMS source to air is reasonably good (1 % - 3 %) compared, for example, to a prior art piezo-active crystal. Piezoactive plastics are almost as good as MEMS. By adding a gas converter according to the present invention, the efficiency could be raised to well over 10 % and the signal-noise ratio can thus be improved significantly. For example, the flow measurement of gases by means of ultrasound would be made significantly more accurate, because we could manufacture a sensor with a broader band and greater phase stability.

2.3 Gas sensor

The sensor can be used to measure both the speed and damping of sound in a gas. The gas can be determined based on these. Of course, the speed and damping of sound also depend on temperature and humidity, so that an accurate measurement requires the measurement of the said variables. In the case of a gas sensor, the resolution of the sensor is not a bottleneck, but the important errors arise from mechanical stress, temperature compensation, etc. Of course, it is important to note that the sensor does not provide unequivocal information on the composition of a gas. If, however, we can use the sensor at higher frequencies too, we can use the changes in the speed and damping of sound to obtain additional information on the composition of the gas. It is generally known that, by measuring both the speed and damping of sound over a very broad frequency range, the humidity of the air, for example, can be determined separately. Depending on the power used, the gas in the cavity of the sensor will heat up (1 - 3 C°). If the gas is very humid, the losses of the cavity will increase and the heating power will increase to evaporate off the moisture. On the other hand, ultrasound reduces the surface tension of the droplets, which then gasify and do not remain on the surfaces. In other words, the method makes it possible to affect the humidity of the gas in the cavity. Because the sensor is not particularly selective, it is most suitable for the detection of an individual gas. It could be, for example, an automobile carbon-dioxide sensor - in the future perhaps a hydrogen sensor, if the use of hydrogen becomes widespread in automobiles - or as a gas detector in buildings or industry. The sensor is particularly suitable as a leak sensor in valves. This is because a valve controls a known gas and the sensor can be located close to the valve in a closed space, where the content of the gas prior to the leak is nearly unchanging. In this application, simultaneous temperature measurement will be sufficient to compensate for errors. Of course, the sensor can act as part of a sensor system, in which a few gases are measured specifically and a missing gas can be detected using the method described. In a gas sensor, the upper diaphragm can be made very stiff and heavy and thus the ultrasound can be made to reflect well from the under surface of the diaphragm. For example, a SOI-based cavity is highly suitable for this purpose. In connection with a gas sensor, we use the name the UtraGas or the UtraGas Sensor.

2.4 Pressure sensor

If the sensor is used for measuring pressure, the cavity should be filled with a known gas. Preferably with a gas with low sound damping and which causes a small effective capacitance and large inductance. Of course, the gas should also be selected so that it will not easily react with silicon or silicon oxide, and will not easily diffuse out of the cavity. A problem with existing MEMS pressure sensors is that the spring constant must be dimensioned in such a way that the height of the cavity will be sufficiently stiff for the maximum pressure not to cause excessive deflection in the diaphragm. The height of the cavity must be kept low, so that capacitive measurement can be made using a low voltage, while the resolution nevertheless remains sufficient. In the UtraSensor this problem does not appear. In other words, by dimensioning the diaphragm to be slack, we increase sensitivity and thus make the device less sensitive to, for example, problems arising from torsion in the substrate. Of course, the downside is that the UltraSensor's effective spring constant depends to some extent on the pressure, according to the following equation h in which h is the height of the cavity, p the pressure, and A the cross-sectional area of the cavity. By substituting typical values in the equation, it will be noticed that, when the height is 100 um or greater, the mechanical spring constant determines the effective spring constant. Thus, the pressure dependence of the spring constant does not cause any great problem. The UtraSensor's great dynamics distinguishes it essentially from present capacitive MEMS sensors, in which the distance of the electrodes from each other is only a few micrometres. In a construction of this kind, so-called capacitive non-linearity arises, which, of course, can be improved by means of feedback. However, feedback often requires a high DC voltage, which is difficult to produce and which causes charging of the surfaces and thus instability. Non-linearity is also caused by the non-linearity of the spring. We call the pressure sensor the UtraPress.

Though the present MEMS pressure sensor is reasonably good, the UtraSensor can provide a cheaper and more stable MEMS sensor for many applications. It is cheap because the increase in sensitivity allows us to fabricate a smaller sensor and, on the other hand, the sensor does not require a special packaging technique, because we measure low impedance, the IC need not be located close to the sensor. Its dynamics and linearity are also superior to present capacitive MEMS sensors. In the case of a pressure sensor, we also do not analyze its sensitivity, as in any case this is sufficient in practical applications. If the sensitivity is not sufficient, we can also use the construction according to Figure lc to improve the sensitivity. In Figure lc, a second cavity, the reflection cavity 6, which is connected to the measurement subject with the aid of holes 5 made in the reflective cover 7, is formed above the sensor cavity 4. This construction is used to eliminate the ultrasound penetrating the sensor structure to the external air space, which has penetrated the diaphragm 3. Measurement inaccuracies are determined in the pressure sensor mainly by systematic errors and problems caused by the package.

2.5 Magnetometer.

If we manufacture a coil, in which either direct current or alternating current travels, on top of the diaphragm, we make the diaphragm to either move or oscillate as a function of the external magnetic field. In the case of direct current, the device operates in exactly the same way as a pressure or gas sensor. We measure the impedance of the change. If, on the other hand, we use alternating current and we tune the upper diaphragm to resonate with this frequency, we create impedance modulation, which appears as sidebands of the carrier wave. If we make a very open and breathing diaphragm, we can make the Q value of the low- frequency resonance very large, and in this way we can maximize the sensitivity of the device. The magnetometer can be combined with a gas or pressure sensor. A problem in the sensor is that, in a simple form it measures only one component of the magnetic field. Due to the gas damping, the dissipation, and through it noise, is greater than that of a traditional MEMS sensor, but on the other hand the spring constant can be made very small, and in this way sensitivity can be maximized. We reckon that the UtraSensor can be manufactured as a product with a sensitivity that can compete with that of present magnetometers. However, because its impedance is low, the electronics can be taken far from the magnetometer. This permits its use in difficult conditions, where electronics cannot be kept close to the measurement object. It can be used, for example, in fusion power plants. The same is true of the magnetometer as of the pressure sensor: it has a wide dynamic range. A magnetometer can also be made by making the upper diaphragm of a magnetic material. We call this invention the UtraMag.

2.6 Acceleration and vibration sensor If the structure breathes sufficiently, and if the mass of the upper diaphragm is made large, we can use the method to build an acceleration sensor. With the aid of the spring constant and the mass, we can adjust the sensitivity and upper-limit frequency of the sensor. Because the spring constant can be very weak, we can manufacture a sensor that is considerably more sensitive than present acceleration sensors. The sensitivity can be calculated from Equation 3 in the present publication. The motion is determined from the equation kx = ma, in which a is acceleration. The damping of the upper diaphragm can be regulated using perforation of the diaphragm. Generally, an acceleration sensor is dimensioned to be over-damped.

The greatest benefit of the acceleration sensor is the wide dynamic range of the UtraSensor. In many measurement subjects, there are a considerable number of vibrations of different frequencies and amplitudes, or high frequency 'squeaking'. Due to the capacitive non-linearity, distortion and intermodulation distortion (the mixing of different frequency components with each other) appear in present MEMS sensors. These cannot be eliminated afterwards. Of course, non-linearity also causes foldover in the noise. This cannot be eliminated afterwards with post-filtering or computation. Figure 3 shows an acceleration or vibration sensor with high resolution and wide dynamics. In the solution, the under surface 3' of the mass 32 acts as the primary sensor element. The closed construction presents moisture affecting the acceleration sensor and the upper stiff diaphragm 3 reflects the power back to the cavity 4, thus increasing the sensitivity of the sensor. The sensor is also not sensitive to pressure variations. By using several ultrasound transmitters and suitably patterning the mass 32, we can also measure lateral movements, i.e. make an x and y-direction

acceleration sensor. In this case, it is preferable for the upper cavity to be the length of one quarter of a wave. We call this sensor the UtraVib. Figure 3 also shows, in a simplified form, the electronics suitable for the invention, in which an alternating electricity source 35 is used between the electrodes 30 of the ultrasound transmitter 2 and, if necessary, direct- voltage biasing 36 in parallel or series with this. The variable being measured is obtained by impedance measurement using the block 34 between the electrodes 30, this being shown in greater detail in Figure 6. Thus, block 34 is used to measure the impedance of the ultrasound transmitter 2, with the aid of its input current and voltage.

2.7 Microphone.

Perhaps the most important application of the UtraSensor is a microphone. This is because it provides a considerably more robust way than present ones of making a precise measurement of the movement of the diaphragm. On the other hand, it is entirely MEMS-based and thus permits a cheap fabrication technique. In this connection, we call the sensor the UtraMic.

It is assumed that the connection of the ultrasound sensor to the sound resonator is so large that the impedance is determined by the effective coil of the resonator, the capacitor, and the losses of the cavity. If we use phase measurement to measure the effective capacitance of the resonance circuit, we can write the resolution in the following form.

Here k is the Boltzmann constant, Q the quality factor of the resonance circuit, the gas temperature, ^ω the angular frequency of the ultrasound, ^{ω =} 271/ , in which/ is the frequency), and U the measuring frequency used in the measurement. It should be noted that C is the effective capacitance, which depicts the ultrasound energy stored in the gas. The effective capacitance of the ultrasound transmitter is determined, biased by direct voltage, from the equation Γ (2) 2^c°o y in which ^ωο is the mechanical resonance frequency, k the spring constant, Co the electrical capacitance, and UQ the bias voltage. The inductance is correspondingly obtained from the equation

in which m is the mass of the diaphragm. The spring constant can naturally be depicted with the aid of the mass and resonance frequency, as follows. k = (2nf m (2)

The movement of the diaphragm is directly proportional to the pressure and the effective surface area of the diaphragm. We can now write the pressure in the form of the resolution

If we operate in a resonance, ^{x ~} λ/2 = ν/2/ , in which v is the speed of sound in air (v = 340 m/s), we can now write the pressure resolution in the form

in which ^ is the density of the cavity's diaphragm and h is its thickness.

It will be noted from the equation, that the resolution can be adjusted mainly by means of the volume of the sound (in this case U), the surface area, and by making the diaphragm as thin as possible. The angular frequency of the ultrasound and the input of the resonance are constant, because the damping of the air per unit of length is directly proportional to the frequency. The effective capacitance C is proportional to the surface area of the cavity, so that the width of the cavity is a linear way to adjust the sensitivity of the microphone. If the width must be made greater than the wavelength of the ultrasound, it is better to divide the ultrasound transmitter into several cavities, which will facilitate the dimensioning of the transmitter. Because the width of the microphone's band must be about 20 kHz, we can only dimension it to be as close as possible to the required band. In this case, the noise temperature T is about 300 K, because the amplifier can be easily designed in such a way that its noise temperature is well below room temperature. We can also write the equation with the aid of the density ^ and thickness h of the diaphragm, as stated in the equation. If we assume that ®Q= 500xl0⁵,/= 10 kHz, C = 1.0 pF, h = 1 μιη, Ή = 5xl0² kg/m³, and U = 3V, m = we obtain 30 μP/Sqrt(Hz) as the pressure resolution. This is sufficient for a practical microphone. Of course, this is an ideal situation, in which the sound resonator by itself determines the losses of the entire construction. If we assume that the impedance of the ultrasound resonator is 3 kΩ, the power consumption will be about 1 mW, which is reasonably large for a microphone, but acceptable.

In an ideal situation, the resolution of the microphone is sufficient, but in practice we may have to widen surface area of the cavity to be greater than the wavelength, in order to increase sensitivity. However, we may have to use the construction shown in Figure lc, to reduce ultrasound losses. Though the UltraMic appears to permit a sufficiently sensitive microphone, optimization must be performed very carefully. After optimization, the equation states that only an increase in the width of the cavity, or an increase in voltage will permit a more sensitive microphone to be made. We have presented a method, with the aid of which changes in sound pressure can be measured at a very high resolution, using a micromechanical ultrasound transmitter. The advantages of the solution are its small size, the independence of the electronics from the sensor, and moisture tolerance. Further, the structure also need not be protected against adhesion of the diaphragm and can be placed directly on a circuit card without a separate package. The challenges of the construction are the adjustment of the manufacturing parameters, the temperature dependence of the sensitivity, and increased power consumption. Of the applications of the invention, the microphone UtraMic is highly reminiscent of an optical microphone, in which the location of the diaphragm is measured optically. In the sensor and method according to the invention there is, however, no need for optics, so that thus there are no surfaces sensitive to dirtying. Another challenge in the invention is how to design the upper diaphragm in such a way that it does not radiate ultrasound to the environment, thus reducing the quality factor of the cavity. This can, however, be avoided by placing a new breathable construction on top of the upper diaphragm, which reflects the power back to the cavity, according to Figure 1 c. The diaphragm 7 of this additional construction can be made so stiff that it cannot radiate ultrasound power out. Of course, this construction increases costs and thickens the component to some extent. This solution should be selected only if the resolution will otherwise not be sufficiently high.

2.8 Combination sensors

Aforementioned combinations can be made from the UtraSensor. First, because the electronics need not be placed next to each sensor, we can make several sensors on one or separate silicon bases. The sensors can be read by, for example, utilizing multiplexing. Specific sensors can also be combined in such a way that the same sensor reads two variables. A low- frequency sensor (pressure, gas, magnetic- field sensor) can be combined with a high-frequency sensor (microphone, vibration sensor, and high-frequency acceleration sensor). For example, if, in the sensor according to Figure lc, the upper diaphragm 7 too were to be flexible, and the intermediate diaphragm 3 reflective but also partly able to be penetrated by ultrasound, we could make the upper diaphragm 7 a pressure sensor and the middle diaphragm 3 an acceleration sensor. A gas sensor could be combined, for example, with the acceleration sensor, etc. In other words, if several different measurement variables are required in the same place, we can reduce the price of the entire sensor module by suitably combining sensors.

3. Electronics The electronics are described with reference to Figure 6, which shows the module 34 of Figure 3 in greater detail. Because a micromechanical ultrasound transmitter 2 is used, it is possible to either bias the transmitter with a DC voltage 36 and create sound at the frequency/ with the aid of an oscillator 35, or we can omit the DC biasing and create sound by using the frequency fll. This is because the power is proportional to the square of the voltage. This is advantageous, because we can effectively eliminate crosstalk. In the measurement of the base impedance, we may need a bridge circuit, or compensation, in order that sufficient power amplification can be taken from the first amplifier stage. Because the frequency of the sound resonator depends on the temperature, the frequency should be adjustable. This means that we use an oscillator 35 controlled by voltage, which is locked with the aid of a phase-sensitive detector, in such a way that the impedance is real. In the case of a microphone, it is advantageous to make the regulation very slow (app. 10 Hz), in which case the signal proportional to the imaginary part of the phase-sensitive detector will be directly proportional to the location of the diaphragm at frequencies 10 Hz - 20 kHz. In the case of a gas sensor and a pressure sensor, the regulating voltage required will tell the value of the variable being measured. In the case of a magnetometer, we can use the regulating voltage directly, or, if alternating current travels in the coil, we make the regulator slower than the frequency of the alternating current, when the magnetic field will be directly proportional to the amplitude of the modulation created.

Thus, according to Figure 6, the signal from the MEMS circuit 1 is led to the amplifier 60 of the circuit 34, the output of which is detected using phase-sensitive detectors 62 and 63. Both the component in phase and the component in a 90-degree phase shift are detected (quadrature detection) from the output. The component of the second branch of the phase detector is rotated through 90 degrees by a phase inverter 61. The signal obtained from the detectors 62 and 63 is processed by the control electronics 65. An analog output 66, which is led to the signal digitalization block 67, is also obtained from the control electronics 65. A digital output 68 is obtained from this circuit. Feedback from the frequency control 69 to the oscillator 35 is

implemented from the control electronics 65. An essential feature of the invention is that the measurement is normal measurement of the real and imaginary parts of a 300 Ohm - 10 kOhm impedance, which can be easily implemented using a simple and cheap IC circuit. Precise measurement of temperature is essential only in the case of a pressure sensor. In a microphone, the dependence of the sensitivity on the temperature should be compensated. A gas sensor too will require temperature measurement and, in precise measurement, possibly also measurement of humidity. If we use fl2 running, we can use DC voltage to adjust the spring constant of the ultrasound sensor. This may be necessary, if the ultrasound transmitter's own resonance differs significantly from the sound resonator's own resonance. Too great a difference may lead to the resonances disintegrating into two separate resonances and in this way the regulation of the frequency may not necessarily function. The resolution will also be significantly reduced.

The parallel capacitance can be eliminated by placing the coil in series or parallel. If the coil is in series, the voltage over the ultrasound transmitter increases and we can increase the power of the ultrasound transmitter without using a high voltage. If we use a high current or voltage, we must make either a passive or an active bridge, in order to compensate the signal being measured. In active compensation, a signal with the opposite phase to the signal going to the sensor is summed with the signal coming from the sensor.

4. Fabrication

Figure 4 shows a way to fabricate one component according to prior art (EP- application 08856883). The ultrasound transmitter 2 is based on a SOI disc 40, the silicon-oxide structure of which is sacrificed in locally, through holes 31 made by RIE etching the disc. The holes 31 are finally filled with polysilicon. The cavity of the transmitter 2 is in a vacuum, so that the ultrasound losses will be determined only from the air, or in this case from the losses of the sound cavity 4. In this case, the electrical connection of the ultrasound resonator is made by growing a polysilicon column 30 in holes made by RIE etching. Before the column 30 is grown, the silicon walls are oxidized. The sound cavity 4 can be made in the disc structure, which is formed of a polysilicon layer 41 and a diaphragm 3. The layer 41 can also be of silicon nitride or amorphic metal. The cavity 4 can be made, for example, by RIE etching the structure 41, in which case the etching stops at the upper diaphragm 3. The discs 40 and 41, 3 can be joined together by fusion bonding. Finally, the disc is cut off, for example, by sawing. If a breather hole is required, it can be made in the upper diaphragm 3, the lower surface of the sound cavity, or the upper surface 9 of the ultrasound transmitter. The size of the breather hole determines lower-limit frequency and should be very low, especially in a microphone. In a high-speed gas sensor, it should be sufficiently large, in order to achieve a fast response time.

Figure 5 shows a sensor, in which the ultrasound cavity 4 is formed by bonding two SOI discs 40 and 80 to each other. A cavity 4 is etched into the upper SOI disc 80. This cover silicon is then polished and thinning grooves 81 are made in it, to permit the diaphragm 3 to make a piston-like movement. Thus, the upper silicon 3 in the upper SOI disc 80 forms a moving measuring diaphragm and the lower 41 a gas cavity 4. According to Figure 7, the sensor according to the invention is based on a surface micromechanical construction, in which the basic structure is a silicon disc 72. In the construction, a silicon-oxide layer 70 is grown on top of the silicon disc and a thin nitride diaphragm 9, in which holes 31 are etched, is grown on top of it. Porous polysilicon 71 is grown on top of the nitride. Silicon oxide 70 is etched through the porous polysilicon 71, which creates a cavity 2. After this, the porous polysilicon 71 is filled by growing additional polysilicon. The pressure of the cavity of the transmitter 2 can be determined at this stage. The spring constant of the diaphragm 9 is determined by the tensile stress of the nitride and can be adjusted by altering the process parameters. An electrical connection is made by etching a hole through the SOI disc 40 and growing an insulating oxide and conductive polysilicon in the hole. The sound cavity 4 is made in the same way as in the SOI disc example of Figure 3. The construction can be manufactured in many different ways. For example, the ultrasound transmitter can be made by growing a polysilicon diaphragm on top of silicon and silicon oxide, and then etching away the silicon oxide from a specific area through holes made in the polysilicon. This permits the manufacture of a low-mass ultrasound transmitter, in which case we can more easily make the cavity determine the effective losses of the entire construction. On the other hand, the ultrasound sensor can be formed of several, e.g., hexagonal transmitters, whereby we can adjust a suitable operating frequency by the size of an individual element and determine the surface area of the cavity. This type of construction is advantageous also because it permits a construction, in which the frequency is high and the cavity wide, and thus the upper diaphragm will not so easily radiate the ultrasound away through the diaphragm of the cavity, thus reducing the resolution.

Figure 3 shows an acceleration sensor, in which the mass is especially thick. In this example, a very high frequency (e.g. 2 Mhz) is used, so that the ultrasound cavity can be very low and an upper cavity is not required. In the sensor in question, it is, of course, sensible to divide the transmitter into several parts (not shown in the figure).

The accompanying description of the invention has presented only some

characteristic features of the invention. It is obvious that embodiments of the invention can vary even widely within the scope of the set of claims.

Thus, in the invention the interaction of an ultrasound transmitter and a cavity connected to it is measured. For its part, the cavity is sensitive to the variable being measured, through its passive sensor diaphragm 9, or the properties of the gas contained in the cavity.

According to the invention, the cavity 4 can be part of a device case, for example, in a pressure-sensor application part of the glass of a diver's watch can be the element transmitting the pressure. Correspondingly, in microphone applications the cavity 4 can be part of the shell of a telephone, etc. Such a form of casing can also be possible, in which an ultrasound circuit according to the invention is on a circuit card and the cavity is formed of a case, which is attached to the circuit card.

The method in accordance with the invention is described in the following with reference to figures 8-22:

The invention is studied first with help of figures 8-15 with two models where Model 1 represents situation where the vibrating membrane (elements 101 and 102 of figure 21) of the sensor is a cantilever on top of which is the piezo.

Model 2 represents situation where the vibrating membrane (elements 101 and 102 of figure 21) is square in shape.

The following assumptions are made referring to figure 9:

Both membrane 102 and piezo 104 have no stiffness in bending modes.

They show stiffness only in the longitudinal strain.

Membrane 102 is elastically isotropic.

The change in the width of the piezo 104 (and membrane) is caused by the Poisson ratio of the piezo 104.

The deflected membrane 102 is parabolic in shape.

The deflection amplitude is very small compared to the membrane width.

Referring to figure 8 membrane deflection is modeled: Model for the membrane deflection

Shape of membrane deflection can be modeled by the following equations: y = Α - L·² The length of membrane can be modeled by the following equations:

Figures 9 and 10 show simplified models of the ultrasound resonator and it will be analyzed in the following:

Two models 1 and 2 are analyzed.

First model 1 represents an isotropic piezo on cantilever

Change in thickness of piezo can be presented by equation

At_p = dU

Lateral stress in piezo can be presented by equation

Lateral stress in cantilever can be presented by equation o_m = c_mAw

Force balance can be presented by equation σ _mt_mh +a _pt_ph = 0

Equilibrium change in cantilever length can be presented by equation

where

d piezoelectric coefficient

U driving voltage

v Poisson ratio

c_p, c_m elastic stiffnesses

h width of the bridge in z direction

Aw change in piezo width

Model 2 represents a square membrane and anisotropic piezo

This analysis is in the following:

Stress in piezo can be presented by equation

Stiffness and piezoelectricity can be presented by equation

Isotropic Si membrane with plane compliance can be presented by equation

2μ + 3λ . .. , _n

c = - ^—— = 252.4 GPa

1 +

2μ Force balance can be presented by equation <5 _m+<5 t = 0 , σ (ζ direction ) = 0 where

e piezoelectric coefficient

E electric field

Cij elastic stiffnesses of piezo

λ, μ Lame elastic constants of Si

Aw change in piezo width Deflection of membrane

Change in cantilever length (model 1) can be presented by equation c_pvw_pdU

Aw - c t + c t

Strain in membrane (model 2) can be presented by equation

and in piezo ε

-5-/. + c + C —2-

Change in membrane width (model 2) can be presented by equation Aw = zw

Deflection of parabolic membrane can be presented by equation

A = _Λ — w Aw

In the following some example values for real solutions: Piezo layer (model 1)

Elastic stiffness c_p= 345 GPa (AIN)

Piezo coefficient d= 4 pm/V (AIN)

Poisson ratio v= 0.287 (AIN)

Thickness t_p= varied 100 - 3000 nm

Width w_p= varied 1 - 30 μιη

Membrane

Elastic stiffness c_m= 167.4 GPa (Si)

Thickness t_m= varied 0.1 - 5 μιη

Width w_m= varied 10 - 500 com

Driving electric field E= U/t_p = varied 0.1 - 5 V/μιη

No resonance effects included

Model 2

cl 1=253.6, cl2= 115.1, cl3=cl2, c33=cl l, c44= 119.3 GPa

el5= -0.6, e31= -0.5, e33= 1.55 As/m2

1= 65, m= 80 GPa

In figure 11 is presented a graph of effect of piezo width in a sensor with 1 micron thick AIN, 1 micron thick Si membrane 102, 100 microns wide membrane, 1 V/μιη drive voltage with no resonance. In figure 12 is presented a graph of Effect of piezo thickness in a sensor with 10 micron wide piezo, 1 micron thick Si membrane 102, 100 microns wide membrane 1 V/μιη drive voltage, and no resonance.

In figure 13 is presented a graph of effect of membrane thickness in a sensor with 1 micron thick AIN, 10 microns wide piezo, 100 microns wide membrane 102, 1 V/μιη drive voltage and no resonance in model 1. In figure 14 is presented graph of effect of membrane size in a sensor with 1 micron thick AIN, 10 microns wide piezo, 1 micron thick Si membrane 102, 1 V/μιη drive voltage and no resonance in model 1. In figure 15 is presented a graph of effect of drive voltage in a sensor with 1 micron thick AIN, 10 microns wide piezo, 1 micron thick Si membrane 102, 100 micron wide membrane and no resonance in model 1. As can be seen from the graph, there is some nonlinearity between the amplitude and the driving voltage. This nonlinearity may be reduced by a bias voltage between the peizo electrodes 103 and 105, typically in a range of 0,5-5 volts, typically 1-2 volts. Alternatively the membrane 102 connected to the piezo structure 104 may be produced such that it has a mechanical pretensioning such that it is deflected to one direction.

In accordance with figures 16 - 21 the manufacturing process is the following:

In figure 16 on silicon wafer 100 is formed an etchstop layer 101. In figure 17 over the etchstop layer 101 is formed a polysilicon layer 102. This layer is typically 1000 nm thick. Over the polysilicon layer 102 is formed a layer of molybdenum 103 forming bottom electrode for a piezo lelectric structure. A piezoelectric layer 104 forming the piezo lelectric structure is formed on the molybdenum 103 layer and partially on the polysilicon layer 102 in accordance with figure 18. The piezoelectric layer is typically 1000 nm thick. In accordance with figure 19 top electrode structure 105 of e.g, aluminium is formed over the piezo layer 104 and polysilicon layer. In accordance with figure 21 the silicon wafer 100 is thinned and a cavity 4 for the ultrasound resonator is formed. The electrode structures 103 and 105 together with pietzoelectric layer 104 form an ultrasonic driving structure controllable by voltage fed between the electrode structures 103 and 105. In accordance with figure 21 etched silicon wafer 106 is bonded to the ultrasound wafer 100. The membrane 107 forms the sensing surface of sensor. The piezo structure 103, 104 and 105 of figures 21 and 22 forms the excitation signal of the sensor structure and the actual signal is formed from the interaction between the ultrasound transmitter 103, 104, 105 and the cavity 4, where the membrane 107 forms a part of the cavity 4 resonator.

Figure 22 shows one embodiment where the piezo structure 104 is formed as a matrix structure of multiple individual elements. Typically, if the invention is used as a micropohone, the number of the individual matrix elements is in a range of 10* 10 - 20*20 elements, in other words comprising 100 - 400 piezo elements. The piezo layer 104 may also be a uniform layer.

Possible materials for the piezo layer are following:

A1N aluminum nitride

PZT lead zirconate titanate

ZnO zinc oxide

BST barium strontium titanate

ScAIN scandiun aluminum nitride

KNN potassium sodium niobate

NBT sodium bismuth titanate

KBT potassium bismuth titanate

LBT lithium bismuth titanate

SBT strontium bismuth titanate

BBT barium bismuth titanate

LN lithium niobate

LT lithium tantalate

etc.

Possible deposition methods to form the piezo matrix or layer 104 are the following: sputtering

evaporation sol-gel screen printing

PLD pulsed laser deposition layer transfer

etc.

Claims

1. Method for manufacturing an ultrasound sensor (1) comprising an ultrasound transmitter (102, 103, 104),

a cavity (4) arranged in connection with this, which is in resonance mode at the ultrasound frequency used, and

- the sensor (1) comprises a passive sensor element (3, 3'), the distance of which from the ultrasound transmitter (2) is selected in such a way that the resonance condition is met, located at the opposite end of the cavity (4) to the ultrasound transmitter (102, 103, 104), characterized in that the ultrasound transmitter (102, 103, 104) is formed as a piezo-element.

2. A method in accordance with claim 1, characterized in that the ultrasound transmitter (102, 103, 104) is formed as a matrix.

3. A method in accordance with claim 1 or 2, characterized in that the ultrasound transmitter (102, 103, 104) is formed as a 10* 10 - 20*20 matrix.

4. A method in accordance with claim 1 , 2 or 3, characterized in that the ultrasound transmitter (102, 103, 104) is formed over a polysilicon layer (102), over which is formed a layer of molybdenum (103) forming bottom electrode, over the layer of molybdenum (103) is formed a piezoelectric structure (104) and over this structure is formed a top electrode structure (105) of aluminium.

5. A method in accordance with any previous claim, characterized in that the piezo layer (104) is made by one of the following methods: sputtering, evaporation, sol-gel, screen printing, PLD pulsed laser deposition, layer transfer.

6. A method in accordance with any previous claim, characterized in that the piezo material is selected from the list of AIN aluminum nitride, PZT lead zirconate titanate ZnO zinc oxide, BST barium strontium titanate, ScAIN scandiun aluminum nitride, KNN potassium sodium niobate, NBT sodium bismuth titanate, KBT potassium bismuth titanate, LBT lithium bismuth titanate, SBT strontium bismuth titanate BBT barium bismuth titanate, LN lithium niobate, LT lithium tantalate

etc.

7. A method in accordance with any previous claim, characterized in that a membrane (102) beneath the piezo layer (104) is formed such that it is mechanically pretensioned.

8. An ultrasound sensor (1) comprising an ultrasound transmitter (102, 103, 104),

- a cavity (4) arranged in connection with this, which is in resonance mode at the ultrasound frequency used, and

- the sensor (1) comprises a passive sensor element (3, 3'), the distance of which from the ultrasound transmitter (2) is selected in such a way that the resonance condition is met, located at the opposite end of the cavity (4) to the ultrasound transmitter (102, 103, 104), characterized in that the ultrasound transmitter (102, 103, 104) is a piezo-element.

9. A sensor in accordance with claim 8, characterized in that the ultrasound transmitter (102, 103, 104) is a matrix.

10. A sensor in accordance with claim 8 or 9, characterized in that the ultrasound transmitter (102, 103, 104) is a matrix in a range of 10* 10 - 20*20.

11. A sensor in accordance with claim 8, 9 or 10, characterized in that the ultrasound transmitter is formed over a polysilicon layer (102), over which is formed a layer of molybdenum (103) forming bottom electrode, over the layer of molybdenum (103) is formed a piezoelectric structure (104) and over this structure is formed a top electrode structure (105) of aluminum.

12. A sensor in accordance with any previous sensor claim, characterized in that the piezo material is selected from the list of A1N aluminum nitride, PZT lead zirconate titanate ZnO zinc oxide, BST barium strontium titanate, ScAIN scandiun aluminum nitride, KNN potassium sodium niobate, NBT sodium bismuth titanate, KBT potassium bismuth titanate, LBT lithium bismuth titanate, SBT strontium bismuth titanate BBT barium bismuth titanate, LN lithium niobate, LT lithium tantalate.

13. A sensor in accordance with any previous sensor claim, characterized in that a membrane (102) beneath the piezo layer (104) is formed such that it is mechanically pretensioned.

14. Use of the sensor defined in any previous sensor clam as a microphone.

15. Method for applying a DC-bias voltage between the electrodes (103, 104) of the sensor structure of any previous sensor clam.

16. a method in accordance with clam 15, characterized in that the voltage is in range 0,5 - 20 V, preferably around 1-2V.