EP3221064B1

EP3221064B1 - Vertical gap actuator for ultrasonic transducers and fabrication of the same

Info

Publication number: EP3221064B1
Application number: EP15784792.2A
Authority: EP
Inventors: Goksen Goksenin YARALIOGLU
Original assignee: Ozyegin Universitesi
Current assignee: Ozyegin Universitesi
Priority date: 2014-11-19
Filing date: 2015-09-09
Publication date: 2018-09-26
Anticipated expiration: 2035-09-09
Also published as: WO2016080931A1; EP3221064A1

Description

TECHNICAL FIELD

The present invention relates generally to the field of ultrasonic transducers. More specifically it is related to electrostatically driven capacitive micromachined ultrasonic transducers (CMUTs).

STATE OF THE ART

Ultrasound was first used for medical diagnosis in 1942 [1, 2] and pulse-echo measurements were first demonstrated in 1950s for tissue characterization [3-5]. Since 1970s, the advances in microelectronics, composite piezoelectric materials [6], and piezoelectric array fabrication technologies resulted in high-resolution ultrasound imaging devices, which have become a widely used tool for medical diagnostics. Piezoelectric materials have dominated the medical ultrasound field. However, despite the wide use, piezoelectric transducers have two important limitations: (1) Bandwidth is limited due to high acoustic impedances of piezoelectric materials. Reduced bandwidth results in poor temporal resolution (therefore poor spatial resolution). (2) The manufacturing process is not compatible with monolithic CMOS integration, which increases the cost and limits the array size and building of complex array geometries. Because the manufacturing technique is limited to cutting (dicing) the piezocomposite block in linear directions, only array elements with rectangular shapes can be produced. To improve the bandwidth of the piezoelectric transducers complex acoustic matching materials are used. This increases the cost of the transducers.
In order to address the limitations of piezoelectric transducers, capacitive micromachined ultrasonic transducers (CMUTs) have been proposed [7]. CMUTs have generated large interest from the MEMS community and offered key performance benefits in terms of bandwidth, scaling to large array sizes, and integration with CMOS compared to its piezoelectric counterparts.
A CMUT transducer is composed of many movable membranes which are actuated by electrostatic forces. These membranes are manufactured in such a way that the cavities at the rear side of them constitute vacuum. The cavity height or the distance between the bottom and the top electrode determines the performance parameters such as the sensitivity, output pressure of the transducer. The membranes are built using micromachining techniques on a highly doped silicon substrate which forms the bottom electrode. There is usually a metal electrode on the membrane. The top metal electrode and the bottom electrode constitute a parallel plate actuator for actuating the membrane. During operation, a high DC electrical voltage is applied between the substrate and the top electrode. Due to the electrostatic force, the membrane is attracted towards the bottom electrode. The induced stress inside the membrane resists the electrostatic force. If an AC voltage is applied on top of the DC voltage, the membrane moves up and down at the same frequency as the AC voltage. This movement of the membrane generates ultrasound in the transducer immersion medium. If the biased membranes are subject to an incoming ultrasound field below the DC voltage, the membrane vibrates at the frequency of the incoming field and an electrical current at the frequency of the incoming field is generated.
The mechanical impedance of the CMUT transducer consisting of membranes is much smaller than that of piezoelectric materials. This enables operation of CMUT transducers with much larger bandwidth. In addition, since the lithography defines the membrane shape and the geometry of array elements, much complex array geometries such as rings, annular arrays can be fabricated. Furthermore, potential of integration with CMOS electronics enables new imaging modalities as well as cost reduction. Since the first introduction of CMUTs, extensive research has been performed on fabrication, modeling and applications. Initially sacrificial release process has been used for fabrication [8, 9]. Later, wafer bonding technique has been introduced [10]. Variations of these two techniques and other fabrication approaches have been reviewed in [11]. An equivalent circuit based approach has been used for modeling of device parameters [12-14]. FEM methods resulted in more accurate device characterization and revealed sources of cross talk [15-17]. CMUTs have also been demonstrated for high frequency applications such as intravascular imaging [18] and 3D volumetric imaging [19]. The literature on the CMUTs is immense, only a few of them have been cited above to show the variety of the work that has been performed up to date.
However, number of commercial ultrasound imaging systems based on CMUTs has been limited despite the large research effort in this area during the last 20 years. Main challenges of CMUTs can be traced to lack of high sound pressure generation, low receive sensitivity and highly nonlinear behavior of the parallel plate actuation.
In CMUT transducers output pressure and receive sensitivity have conflicting requirements over the cavity height (gap). The output pressure is fundamentally limited by the cavity height (distance between the bottom surface of the membrane and the top surface of the substrate). The membrane cannot travel more than the cavity height. Therefore, to be able to generate high output pressure levels one needs to build transducers with large cavity heights. On the other hand, high receive sensitivity requires high e-fields at small cavity heights. This can be achieved by reducing the cavity height. Therefore, high output pressure and high receive sensitivity cannot be obtained at the same time from the same transducer geometry.
CMUT transducers also generate high level of non-linearity due to the parallel plate actuation. As the membrane displaces when the electrical field inside the gap changes, this results in spring softening effect and shifts the resonant frequency. As a result, high levels of non-linearity are generated.
To solve above problems various solutions have been proposed in the literature. Using piston-shaped membranes have been proposed to achieve piston like membrane motion to increase output pressure [23, 24]. Patent document US2007/0058825 discloses a capacitive transducer wherein the center portion of the diaphragm is increased in thickness in comparison with the near-end portion. Collapse-snap back operation [25] and collapse mode [26, 27] have been introduced also to increase the output pressure. However, these approaches had marginally increased the output pressure and they also suffered from instabilities due to electric breakdown (break-down) issues. Double electrode configuration has been also proposed to separate transmit and receive signal paths [28]. With this approach, two electrodes were placed onto the membrane. From these, the electrode provided at the center of the membrane is used to collect ultrasonic signals. The electrode around the central electrode, due to a bias voltage applied on it, bends the membrane toward the substrate and reduces the gap between the central electrode and the substrate. This helped improving receive sensitivity, but high electrical cross talk between the receive and the transmit paths and increased number of connections resulted in more complex electronics limiting the usability of the approach. In another approach, Ying proposed adding a flat electrode which moves like a piston [29] in the CMUT cavity. In their publication they did not report the output pressure but they reported only 70% increase in the receive voltage. The disadvantage of this method is the increased complexity of the fabrication methods. Huang proposed using a comb drive [30]. This structure can potentially increase the output pressure but the method does not address building very small vertical gaps and does not propose a solution for separating the requirements for maximum output pressure and increased sensitivity. In addition, building multiple fingers as described in Huang's invention [30] is difficult to achieve.
None of the above approaches addressed the fundamental problem which is the conflicting requirement of receive sensitivity and maximum output pressure on the cavity height, constituting the reason for the low performance of the CMUTS. There is a need for a new actuation method for the CMUT transducers that will separate the dependence of the output pressure and receive sensitivity on the gap.

OBJECT OF THE INVENTION

It is an object of the present invention to provide a new geometry that will solve the low output pressure problem of CMUTs. In this geometry, a thick part in form of a piston is added to the middle part of the membrane. This part will move within a cavity opened on the substrate. The movement of the membrane is achieved by an electric field created on the gap between the piston-shaped part of the membrane and the side walls of the cavity in the substrate. In this geometry, the motion of the membrane does not affect the height of the gap where the actuation forces are built. Also, as the gaps between the thin part of the membrane and the bottom part of the piston and the substrate are selected too large, these do not interfere with the membrane motion. This enables membrane to move without any hard limits. The proposed geometry can be fabricated using micromachining techniques and transducers can be manufactured by repeating the cellular structure.
In order to achieve the aforementioned object, the present invention provides an electromechanical micromachined transducer according to claim 1.
In a preferred embodiment of said electromechanical micromachined transducer, the membrane consists of doped single crystal silicon, doped poly-silicon layer or both.
In a preferred embodiment of said electromechanical micromachined transducer, the insulating layer consists of silicon dioxide.
In a preferred embodiment of said electromechanical micromachined transducer, the substrate consists of doped single crystal silicon.
In a preferred embodiment of said electromechanical micromachined transducer, the membranes have an electrically connected structure to form 1D, 2D or annular arrays. In order to achieve the aforementioned object, the present invention provides an electromechanical micromachined transducer production method according to claim 9.
Structural and characteristic properties and all advantages of the invention will be understood more clearly due to the following drawings and the detailed disclosure written by referring to these drawings and therefore the evaluation has to be made by taking into account these drawings and the detailed disclosure.

BRIEF DESCRIPTION OF THE DRAWINGS

FIGURE 1 shows the proposed geometry for the vertical gap CMUT cell.
FIGURE 2 shows the electrical field lines between the side walls of the first silicon wafer gap and the piston part of the membrane.
FIGURE 3 shows the geometry of a standard CMUT cell.
FIGURE 4 shows the parallel plate modeling of a standard CMUT cell.
FIGURE 5 shows the equivalent circuit modeling of a standard CMUT cell.
FIGURE 6 shows the simplified modeling of the vertical gap CMUT.
FIGURE 7 shows the equivalent circuit modeling of the vertical gap CMUT.
FIGURE 8.1, FIGURE 8.2, FIGURE 8.3, FIGURE 8.4, FIGURE 8.5, FIGURE 8.6,
FIGURE 8.7,
FIGURE 8.8, FIGURE 8.9, FIGURE 8.10, FIGURE 8.11, FIGURE 8.12, show a method to fabricate vertical gap CMUTs.
FIGURE 9.1, FIGURE 9.2, FIGURE 9.3, FIGURE 9.4, FIGURE 9.5, FIGURE 9.6,
FIGURE 9.7, FIGURE 9.8, FIGURE 9.9, show another method to fabricate vertical gap CMUTs.
FIGURE 10.1, FIGURE 10.2 AND FIGURE 10.3 show 1D, 2D and annular transducer arrays.

It is not necessary to scale the drawings and the details which are not necessary for understanding the present invention can be neglected. Beside this, at least substantially identical elements or at least substantially identical functionalized elements are shown with the same numbers.
Reference List

100.: Membrane
1001.: Side walls of the thick central part

101.: Vertical gap
102.: Substrate
103.: Thick central part
104.: Vertical side walls of the substrate
105.: Cavity
106.: Horizontal gap 1
107.: Horizontal gap 2
108.: Thin peripheral part
109.: Insulating layer
110.: Horizontal component of the electric field
111.: Fringing field

117.: Gap distance
118.: Parallel plate electrode
119.: Spring
120.: Electrical capacitor
121.: Transformer
122.: Negative capacitor
123.: Series capacitor

124.: Inductor
125.: Resistor
126.: Piston

134.: Dry oxide sacrificial layer
135.: Poly-silicon layer
136.: Oxide layer

142.: Device layer
143.: Array elements belonging to 1D array

146.: Array elements belonging to 2D array
147.: Array elements belonging to annular array

C ₀:: Device capacitance
Ṽ_DC :: Voltage
V:: DC Voltage
ε ₀:: Dielectric permittivity of vacuum
A:: Electrode area
-c_s :: Capacitance of the softening of spring constant
c_m :: Capacitance of spring constant
c_soft :: Capacitance corresponding to spring constant together with softening
l_m :: Inductance
ĩ _m:: Current

r:: Resistance corresponding to losses
n:: Transformer ratio
h:: Piston height
g:: Gap distance
k:: Spring constant
r_p :: Membrane radius
x:: Distance

DETAILED DESCRIPTION OF THE INVENTION

The main objective of this invention is to develop novel cell geometry for CMUT transducers where the sensitivity and the maximum output pressure do not have conflicting requirements over the gap. To achieve this, instead of parallel plate actuation, electrodes will be positioned such that the cavity height does not change with the membrane (100) motion. In this completely novel geometry, an electrostatic gap is formed vertically in the capacitor formed by a piston like part of the membrane (100) with the side walls of the gap. This new geometry is shown in Fig.1. In this Figure, the actuation of the membrane (100) is achieved through a vertical gap (101). The membrane (100) is composed of two parts: thick central part (piston part) (103) and thin peripheral part (108). The thin peripheral part (108) is more flexible than the thick central part (103). The thick central part (the piston part) (103) has a vertical side wall (1001) as shown in the Fig.1. When a force applied along the axis of the membrane (100), bending is observed at the thin peripheral part (108) of the membrane (100). The membrane (100) could be in circular, square or rectangular shape. It is made of a highly conducting material, preferably highly doped silicon. To increase the electrical conductivity, a metal electrode (not shown) could be deposited on top of the membrane (100). The substrate (102) which can be preferably silicon wafer is etched to form a cavity (105) with several steps as shown in Fig.1. Cavity (105) is composed of three different gaps: horizontal gap 1 (106) under the thin peripheral (108) part of the membrane (100), horizontal gap 2 (107) under the thick central part (the piston part) (103) of the membrane (100) and vertical gap (101) between the vertical side walls (104) of the substrate (102) and the thick central part side walls (1001). First silicon wafer , namely substrate (102) is also made of a highly conducting material, preferably highly doped silicon. Membrane (100) is attached to the substrate (102) through an insulating material (109) whose electrical impedance is very high. The insulating material is preferably made of silicon dioxide. The thick central part (103) can move freely in the cavity (105) of the substrate (102). When a voltage is applied between the membrane (100) and the substrate (102), electrical field is formed between the membrane (100) and the substrate (102) over the cavity (105).
Fig. 2 shows electric field formed in the gap when there is voltage applied between the membrane (100) and the substrate (102). The vertical gap (101) is substantially smaller than the horizontal gap 1 (106) and the horizontal gap 2 (107). Therefore there is a stronger electrostatic force between the side walls (104) of the substrate (102) and the thick central part sidewalls (1001). This electric field has two components as shown in Fig 2: electric field horizontal component (110) and fringing fields (111). The horizontal component does not move the membrane (100) because it is balanced around the perimeter of the piston. However, fringing fields (111) between the thick central part (103) and the side walls (104) of the substrate (102) in horizontal gaps as shown in Fig. 2 can apply a net force on the thick central part (103), moving it towards the substrate (102). The gaps between the membrane (100) thin peripheral part (108) and the thick central part (103) and the substrate (102) can be chosen substantially larger than the maximum membrane (100) stroke such that they these two gaps do not interfere with the membrane (100) motion. Horizontal gap 1 (106) and horizontal gap 2 (107) can be different from each other and horizontal gap 2 (107) can be chosen much larger than horizontal gap 1 (106).
Fig.3 shows the schematic of a standard CMUT cell. The CMUT is composed of a membrane (100), substrate (102) and an insulator layer (109) which electrically isolates the membrane (100) and the substrate (102). The cavity (105) is obtained by etching into the substrate (102). The bottom of the cavity and the membrane (100) forms a parallel plate actuator. When the membrane (100) is actuated by an applied voltage between the membrane (100) and the substrate (102), the gap distance (117) changes with the applied voltage. The main difference between the geometry constituting the object of the invention disclosed in this patent application and the geometry shown in Fig 3 is that the vertical gap (101) of Fig.1 can be chosen as small as possible for increased transmit and receive sensitivity. The important feature of this design is that the selection of the gap does not impose any limits on the maximum output pressure.
To the first order, the proposed geometry can be analysed using parallel plate theory and equivalent circuit approach. Fig. 4 shows the simplified model for a standard CMUT cell. In Fig. 4 the membrane (100) is modelled by a spring (119) and a parallel plate electrode (118). The electrode area is assumed to be A. The gap distance is denoted by g. The details of the calculation of the parameters for the equivalent circuit have been given in [15, 22] for a standard CMUT cell.
Fig. 5 shows the equivalent circuit model for the standard CMUT cell. The transformer (121) connects electrical domain to mechanical domain. On the electrical side electrical capacitor (120) represents the electrical parallel capacitance of the CMUT. On the mechanical side the membrane (100) impedance is modelled by a series (array) combination of a capacitor (123) and an inductor (124). The spring softening is modelled by another capacitor (122). The radiation losses are modelled by a resistor (125) which is terminated the acoustic port of the transducer.
Fig 6 shows the simplified model for vertical gap CMUT. The membrane (100) is modelled by a spring (119) and a piston (126). This time the electrodes are on the side of the piston. The piston radius, namely the membrane radius is denoted by r_p. h denotes the length of the electrodes. Following the same technique, one can build an equivalent circuit for the vertical gap CMUTs. One such circuit is shown in Fig. 7. Similar to the circuit shown in Fig. 5, the equivalent circuit is composed of an electrical capacitor (120) in the electrical part, a transformer (121), a series capacitor (123) and an inductor (124). The output port is again terminated by a resistor (125). The main difference between this circuit and the circuit shown in Fig 5 is that this circuit does not have any spring softening capacitor. The main reason for the lack of spring softening is that the motion of the membrane (100) does not change the gap distance.
The calculation of the circuit parameters are given in Table 1. Assuming electrical capacitance and the membrane impedance can be made the same for both types of CMUTs, the main difference between the two circuits is the transformer ratio and the spring softening capacitance. The transformer ratio represents the conversion efficiency between electrical and mechanical domains. By using the equations shown in Table 1, one can calculate the ratio of the transformer n's as following: $\frac{n_{vertical}}{n_{st}} = \frac{2 g_{st}^{2}}{r_{p} g_{v}}$

In the above equation, the electrode radius (r_p ) of the standard CMUT is the same as the piston radius of vertical gap CMUT. g_st and g_v represents the gap height for standard and vertical gap CMUTs, respectively. For 10 micro-meter radius and 0.5 micro-meter standard gap CMUT, if one can build a vertical gap device with 50 nm gap, it is possible to get similar transformation ratios. Based on this first order approach, one can achieve similar sensitivities using both devices.

Table 1.First order modeling for the vertical gap CMUT cell.

	Standard CMUT	Vertical gap CMUT
Electrostatic Force	$F_{E} = \frac{1}{2} \frac{A ε_{0}}{{(g - x)}^{2}} V^{2}$	$F_{E} = \frac{1}{2} \frac{2 π r_{p} ε_{0}}{g} V^{2}$
Total Force on the membrane	$F_{N} = \frac{1}{2} \frac{A ε_{0}}{{(g - x)}^{2}} V^{2} - kx$	$F_{N} = \frac{1}{2} \frac{2 π r_{p} ε_{0}}{g} V^{2} - kx$
Small signal spring constant	$\tilde{k} = - \frac{d F_{N}}{dx} = - \frac{A ε_{0}}{{(g - x)}^{3}} V^{2} + k$	$\tilde{k} = - \frac{d F_{N}}{dx} = k$
Spring softening	$\frac{1}{c_{s}} = \frac{A ε_{0}}{{(g - x)}^{3}} V^{2}$	$\frac{1}{c_{s}} = 0$
Electromechanical conversion ratio	$n = \frac{d F_{E}}{dV} = \frac{A ε_{0}}{{(g - x)}^{2}} V$	$n = \frac{d F_{E}}{dV} = \frac{2 π r_{p} ε_{0}}{g} V$
Spring constant	$c_{m} = \frac{1}{k}$	$c_{m} = \frac{1}{k}$

Another, interesting feature of the vertical gap device is that there is no collapse phenomena and spring softening effect. For standard CMUT cells, as the bias voltage increases, the membrane (100) gets closer to the substrate (102). Beyond certain voltage, the membrane (100) collapses onto the substrate (102). The voltage where the membrane (100) collapses is called collapse voltage. This is due to the fact that the electrostatic force gradient increases as the bias voltage increases and after certain voltage, the force gradient exceeds the spring constant of the membrane (100) and the membrane (100) is pulled towards the substrate. For vertical gap devices, because the electrostatic force is independent of membrane displacement (x), the force gradient is always zero. Therefore there is no collapse voltage associated with these devices. For standard CMUT cells, there is also spring softening effect which is due to the fact that electrostatic force gradient is not zero. When the membrane (100) moves in an electric field, the apparent spring constant of the membrane (100) reduces. This is called spring softening effect. For vertical gap CMUTs, because the electric force gradient is zero, there is no spring softening effect. Lack of spring softening effect increases the linearity of the vertical gap device. The relation between the membrane (100) motion and the applied voltage is quadratic; therefore a simple algorithm can be designed to reduce the harmonic content of output pressure from vertical gap CMUTs.
In the following, it is explained how to fabricate the membrane (100) geometry array disclosed in this patent by using two exemplary silicon micromachining methods. The main purpose of both methods is to fabricate a very thin vertical gap (101) between the walls of the gap at the rear side of the membrane (100) and the thick central part (piston part) (103) of the membrane (100). This small vertical gap (101) is achieved by growing a thin dry oxide sacrificial layer (134) and etching this layer from the back side of the wafer.
Fig. 8.1, Fig. 8.2, Fig. 8.3, Fig. 8.4, Fig. 8.5, Fig. 8.6, Fig. 8.7, Fig. 8.8, Fig. 8.9, Fig. 8.10, Fig. 8.11, Fig. 8.12 show a process to build vertical gap CMUT where the vertical gap height can be made substantially small. The fabrication process starts with a highly doped silicon wafer, namely a substrate (102). The conductivity of the starting substrate (102) should be high because this wafer will provide the side wall electrodes required to actuate the membrane (100). The first silicon wafer (102) is etched as shown in Fig 8.2. For this etch first the substrate (102) is coated by a photoresist. Then the photoresist is patterned to define the gaps as shown in Fig 8.2. One can also start with an SOI wafer where the gap depth is determined by the device layer of the SOI (Silicon On Insulator) wafer. This provides more uniform gap depth across the wafer. In the next step (Fig 8.3), a thin sacrificial layer of dry oxide (134) is grown over the wafer surface. This oxide layer will be later etched to define the vertical gap height. The gap height is substantially small, between 50 nm to 200 nm. To define such a small gap height a well-controlled dry oxide deposition is preferred. Fig 8.4 demonstrates poly-silicon layer (135) deposition. LPCVD method can be used to deposit poly-silicon. During poly-silicon deposition, poly-silicon should be doped to achieve high conductivity. In the next step (Fig 8.5), the poly-silicon is etched back down to oxide layer, leaving poly-silicon only in the cavities etched in Fig 8.2. Poly-silicon can be etched using dry etching or chemical wafer polishing (CMP). In the next step, as shown in Fig 8.6, more poly-silicon is deposited and is patterned. This part of the poly-silicon will be part of the piston. In the next step (Fig. 8.7), a thick conformal oxide layer (136) is grown over the wafer surface. This oxide layer will provide the insulation layer between the membrane (100) and the substrate (102). In the subsequent step (Fig. 8.8), the thick oxide layer is patterned as shown in the figure. This also defines the membrane diameter. Before bonding an SOI wafer, the wafer surface is planarized using CMP process (Fig. 8.9). After planarization, an SOI wafer is bonded to the top surface of the wafer (Fig. 8.10). After removing the handle and the box oxide, a membrane (100) is formed. In the next step (Fig. 8.11), the back side of the wafer is etched using deep reactive ion etching down to the oxide layer as shown in Fig 8.11. This step exposes the oxide layer that is filling the vertical gap. At step 12, the vertical side wall is etched until it is completely cleared of oxide. The oxide etching can be done using vapor phase HF or wet BOE (buffered oxide etch). After step 12, the main structure of the membrane (100) is completed. In the subsequent step the bottom side of the wafer can be bonded to another wafer at low pressure to evacuate the air in the back side of the membrane cavity (the reason for naming this wafer second wafer is the evaluation of the substrate (102) as the first wafer). Further metallization steps can be found in CMUT fabrication references [11, 12] and same methods can be employed here.
Fig 9.1, Fig. 9.2, Fig. 9.3, Fig. 9.4, Fig. 9.5, Fig. 9.6, Fig. 9.7, Fig. 9.8 and Fig. 9.9 shows an alternative approach which also starts with a highly doped silicon wafer (102). At the first step (Fig 9.2), the surface of the wafer is etched to form a trench as shown in the figure. This trench will be filled with poly-silicon to form the piston part of the membrane (100) in the subsequent processing steps. At step 3 (Fig 9.3), a thin sacrificial layer of dry oxide (134) is grown on the silicon wafer surface. Dry oxidation is the method of choice for this step due to its high uniformity and well-controlled thickness. At step 4 (Fig 9.4), a thick layer of poly-silicon layer (135) is deposited. At step 5 (Fig 9.5), the poly-silicon layer is patterned and etched using reactive ion etching. In the subsequent step, the wafer surface is covered by a thin layer of oxide (136). At step 6, (Fig 9.6) this oxide layer is patterned and etched. The oxide pattern at this step determines the diameter of the membrane (100). After oxide etch, the wafer is polished using CMP process. The purpose of the polishing is to planarize the wafer surface by bringing the oxide part and the poly-silicon islands to the same height for the subsequent bonding process. At step 7 (Fig 9.7), an SOI wafer is bonded on top of the wafer. The device layer (142) is attached to the oxide layer (136) and the poly-silicon part (135). At step 8 (Fig 9.8), the back side of the wafer is patterned and etched until the thin dry oxide sacrificial layer (134) is exposed using deep reactive ion etching. In the next step (Fig 9.9), the thin oxide layer covering the bottom part of the poly-silicon layer (135) and the vertical gap is etched and the membrane (100) is released. In the subsequent processing steps (not shown) the back side of the wafer is bonded to another wafer at low pressure to seal the gaps and electrodes are fabricated to provide electrical interface to the mechanical structure.
Figure 10.1, Figure 10.2 and Figure 10.3 shows various transducer array geometries where the cells of the transducers are fabricated as explained above. The array geometries consist of electrodes providing electrical connections to the membrane (100) and the substrate (102). The array geometries are not limited to the ones that are shown in this figure. In Fig 10.1, array elements (143) belonging to a linear 1D array are formed by electrically isolating the membranes (100) on the substrate (102). The boundary of the isolation layer defines the membranes (100) which are shown by dotted lines in the figure. Fig 10.2 shows array elements (146) belonging to a 2D array where the membrane layer is patterned to form a 2D array of transducer elements. Fig.10.3 shows an array formed by the array elements (147) belonging to an annular array. In all these configurations, the array geometry is defined by patterning the conducting membrane layer.

REFERENCES

[1] F.V.HUNT, ELECTROACOUSTICS: THE ANALYSIS OF TRANSDUCTION, AND ITS HISTORICAL BACKGROUND. NEW YORK: ACOUSTICAL SOCIETY OF AMERICA, 1982.
[2] DUSSIK, K. T., "POSSIBILITY OF USING MECHANICAL HIGH FREQUENCY VIBRATIONS AS A DIAGNOSTIC AID" Z. NEUROL. PSYCHIAT. 174, 153-168 (1942).
[3] WILD, J., "THE USE OF ULTRASONIC PULSES FOR MEASUREMENT OF BIOLOGICAL TISSUES AND THE DETECTION OF TISSUE DENSITY CHANGES," SURG. 127 (2), 183-188 (1950)
[4] REID, J. AND WILD, J. "APPLICATION OF ECHO RANGING TECHNIQUES IN DETERMINATION OF STRUCTURE OF BIOLOGICAL TISSUES," SCI., 115, 226-230 (1952).
[5] HOWRY, D. AND BLISS, W., "ULTRASONIC VISUALIZATION OF SOFT TISSUE STRUCTURES OF THE BODY," J. LAB. AND CLIN.MED. 40, 579-592 (1952).
[6] SMITH, W. A., "THE ROLE OF PIEZOCOMPOSITES IN ULTRASONIC TRANSDUCERS," IN PROC. IEEE ULTRASON. SYMP., 1989, PP. 755-766.
[7]M. I. HALLER AND B. T. KHURI-YAKUB, "A SURFACE MICROMACHINED ELECTROSTATIC ULTRASONIC AIR TRANSDUCER," IN PROC. IEEE ULTRASON.SYMP., 1994, PP. 1241-1244.
[8] H.T. SOH, I. LADABAUM, A. ATALAR, C.F. QUATE, AND B.T. KHURI-YAKUB, "SILICON MICROMACHINED ULTRASONIC IMMERSION TRANSDUCERS," APPL. PHYS. LETT., VOL. 69, PP. 3674 - 3676, DEC. 1996.
[9] M.I. HALLER, AND B.T. KHURI-YAKUB, "A SURFACE MICROMACHINED ELECTROSTATIC ULTRASONIC AIR TRANSDUCER," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOL.43, NO.1, PP.1-6, JAN 1996.
[10] HUANG, Y., SANLI ERGUN, A., HÆGGSTRÖM, E., BADI, M.H., KHURI-YAKUB, B.T., "FABRICATING CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS WITH WAFER-BONDING TECHNOLOGY," JMEMS 12 (2), PP. 128-137, 2003.
[11] A.S. ERGUN, Y. HUANG, X. ZHUANG, O. ORALKAN, G.G. YARALIOGLU, AND B.T. KHURI-YAKUB, "CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS: FABRICATION TECHNOLOGY," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOL. 52, NO. 12, PP. 2242- 2258, DEC. 2005.
[12] BADI, M.H. ; YARALIOGLU, G.G. ; SANLI ERGUN, A. ; YONGLI HUANG ; KHURI-YAKUB, B.T., "IMPROVED MODELING AND FABRICATION TECHNIQUES FOR CAPACITIVE MICROMACHINED ULTRASONIC LAMB WAVE TRANSDUCERS," ULTRASONICS, 2003 IEEE SYMPOSIUM ON.
[13] LOHFINK, A. ;ECCARDT, P.-C., "LINEAR AND NONLINEAR EQUIVALENT CIRCUIT MODELING OF CMUTS," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 52 , ISSUE: 12, 2163 - 2172, 2005.
[14] OGUZ, H.K. ;ATALAR, A. ; KOYMEN, H., "EQUIVALENT CIRCUIT-BASED ANALYSIS OF CMUT CELL DYNAMICS IN ARRAYS," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 60 , ISSUE: 5, 2013.
[15] YARALIOGLU, G.G. ; ERGUN, A.S. ; BAYRAM, B. ; HAEGGSTROM, E. ; KHURI-YAKUB, BUTRUS T., "CALCULATION AND MEASUREMENT OF ELECTROMECHANICAL COUPLING COEFFICIENT OF CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 50, ISSUE: 4,449 - 456, 2003.
[16] BAYRAM, B. ; KUPNIK, M. ; YARALIOGLU, G.G. ; ORALKAN, O. ; ERGUN, A.S. ; WONG, S.H. ; KHURI-YAKUB, B.T. , "FINITE ELEMENT MODELING AND EXPERIMENTAL CHARACTERIZATION OF CROSSTALK IN 1-D CMUT ARRAYS", ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 54, ISSUE: 2, 418 - 430, 2007.
[17]G.G. YARALIOGLU, S.A. ERGUN, AND B.T. KHURI-YAKUB, "FINITE-ELEMENT ANALYSIS OF CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS," IEEE TRANSACTIONS ON UFFC, VOL. 52, NO. 12, PP. 2185-2198, DEC. 2005.
[18] YEH, D.T. ; ORALKAN, O. ; WYGANT, I.O. ; O'DONNELL, M. ; KHURI-YAKUB, B.T., "3-D ULTRASOUND IMAGING USING A FORWARD-LOOKING CMUT RING ARRAY FOR INTRAVASCULAR/INTRACARDIAC APPLICATIONS," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON VOLUME: 53 , ISSUE: 6, PAGE(S): 1202 - 1211, 2006.
[19] O. ORALKAN, A.S. ERGUN, C.H. CHENG, J.A. JOHNSON, M. KARAMAN, T.H. LEE, AND B.T. KHURI-YAKUB, "VOLUMETRIC ULTRASOUND IMAGING USING 2-D CMUT ARRAYS," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOL. 50, NO. 11, PP. 1581-1594, NOV 2003.
[20] MILLS, D.M., SMITH, L.S., "REAL-TIME IN-VIVO IMAGING WITH CAPACITIVE MICROMACHINED ULTRASOUND TRANSDUCER (CMUT) LINEAR ARRAYS," ULTRASONICS, 2003 IEEE SYMPOSIUM ON, VOLUME: 1, PAGE(S): 568 - 571 VOL. 1, 2003
[21] S. H. WONG, M. KUPNIK, R. D. WATKINS, K. BUTTS-PAULY, AND B. T. KHURI-YAKUB, "CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS FOR THERAPEUTIC ULTRASOUND APPLICATIONS," BIOMEDICAL ENGINEERING, IEEE TRANSACTIONS ON, VOL. 57, NO. 1, PP. 114-123, JAN. 2010.
[22] YARALIOGLU, G.G., BADI, M.H., ERGUN, A.S., KHURI-YAKUB, B.T., "IMPROVED EQUIVALENT CIRCUIT AND FINITE ELEMENT METHOD MODELING OF CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS," PROCEEDINGS OF THE .
[23] HUANG Y. ; ZHUANG X.; HAEGGSTROM, E. ; ERGUN, A. ; CHING-HSIANG CHENG ; KHURI-YAKUB, B., "CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCERS WITH PISTON-SHAPED MEMBRANES: FABRICATION AND EXPERIMENTAL CHARACTERIZATION," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 56 , ISSUE: 1, 136 -145, 2009.
[24] LEE, B.C., NIKOOZADEH, A., PARK, K.-K., KHURI-YAKUB, B.T., "UNDERSTANDING CMUTS WITH SUBSTRATE-EMBEDDED SPRINGS," IEEE INTERNATIONAL ULTRASONICS SYMPOSIUM, IUS , ART. NO. 6293679 , PP. 1008-1011,2011.
[25] BAYRAM, B. ; ORALKAN, O. ; ERGUN, A.S. ; HAEGGSTROM, E. ; YARALIOGLU, G.G. ; KHURI-YAKUB, B.T., "CAPACITIVE MICROMACHINED ULTRASONIC TRANSDUCER DESIGN FOR HIGH POWER TRANSMISSION" VOLUME: 52, ISSUE: 2,, PAGE(S): 326 - 339, 2005.
[26] ORALKAN, O. ; BAYRAM, B. ; YARALIOGLU, G.G. ; ERGUN, A.S. ; KUPNIK, M. ; YEH, D.T. ; WYGANT, I.O. ; KHURI-YAKUB, B.T., "EXPERIMENTAL CHARACTERIZATION OF COLLAPSE-MODE CMUT OPERATION" ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 53 , ISSUE: 8, PAGE(S): 1513 - 1523, 2006.
[27]OLCUM, S. ; YAMANER, F.Y. ; BOZKURT, A. ; KOYMEN, H. ; ATALAR, A., "CMUT ARRAY ELEMENT IN DEEP-COLLAPSE MODE", ULTRASONICS SYMPOSIUM (IUS), 2011 IEEE INTERNATIONAL, PAGE(S): 108 - 111, 2011.
[28] GULDIKEN, R.O. ; MCLEAN, J. ; DEGERTEKIN, F., "CMUTS WITH DUAL ELECTRODE STRUCTURE FOR IMPROVED TRANSMIT AND RECEIVE PERFORMANCE," ULTRASONICS, FERROELECTRICS AND FREQUENCY CONTROL, IEEE TRANSACTIONS ON, VOLUME: 53 , ISSUE: 2, PAGE(S): 483 - 491,2006.
[29] JEONG, B.-G., KIM, D.-K., HONG, S.-W., CHUNG, S.-W., SHIN, H.-J., "PERFORMANCE AND RELIABILITY OF NEW CMUT DESIGN WITH IMPROVED EFFICIENCY," SENSORS AND ACTUATORS, A: PHYSICAL 199 , PP. 325-333, 2013.
[30] US PATENT APPLICATION# 2009/0152980 A1, HUANG , ELECTROSTATIC COMB DRIVER ACTUATOR/TRANSDUCER AND FABRICATION OF THE SAME.

Claims

An electromechanical micromachined transducer having a cellular structure, comprising:
- a free movable membrane (100) consisting of a thin peripheral part (108) and a thick central part (103),

- an insulating layer (109) supporting the membrane (100),

- a first wafer being an incised substrate (102) bound to the membrane by the insulating layer (109),

- a second wafer bound to the substrate (102), wherein the cellular structure is arranged such that it presents:

- vertical gaps (101) each having a gap distance between the side walls (1001) of the thick central part of the membrane (100) and vertical side walls (104) of the incised substrate (102),

- a first horizontal gap (106) having a first gap distance between the membrane thin peripheral part (108) and the incised substrate (102), and

- a second horizontal gap (107) having a second gap distance between the bottom part of the thick central part (103) and the second wafer, characterized in that the first gap distance and the second gap distance of the first and second horizontal gaps (106, 107), respectively, are larger than the gap distance of the vertical gaps (101),

- wherein the membrane (100) and the substrate (102) are conductive, and

- wherein the side walls (1001) of the thick central part of the membrane (100) and vertical side walls (104) of the incised substrate (102) are configured for application of an actuation voltage.
A micromachined transducer according to claim 1, characterised in that the membrane (100) consists of doped single crystal silicon, doped poly-silicon layer (135) or both.
A micromachined transducer according to claim 1, characterised in that the insulating layer (109) consists of silicon dioxide.
A micromachined transducer according to claim 1, characterised in that the substrate (102) consists of doped single crystal silicon.
A micromachined transducer according to claim 1, characterised in that the membranes have an electrically connected structure to form array elements (143) belonging to 1D array, array elements (146) belonging to 2D array or array elements (147) belonging to annular array.
A micromachined transducer according to claim 1, characterised in that the gap distance of the vertical gaps (101) is between 50 nm and 200 nm.
A micromachined transducer according to claim 1, characterised in that the second gap distance of the second horizontal gap (107) is larger than the first gap distance of the first horizontal gap (106).
A micromachined transducer according to claim 1, wherein the cellular structure further comprises a thin oxide layer (134) between the insulating layer (109) and the substrate (102).
An electromechanical micromachined transducer production method, comprising:
- etching a cavity on a doped substrate (102) being a first wafer to form a thick central part (103) of a membrane (100),

- depositing substantially a thin dry oxide sacrificial layer (134) on the substrate (102) to form vertical gaps (101), wherein the vertical gaps (101) are formed between the side walls (1001) of the thick central part of the membrane (100) and vertical side walls (104) of the incised substrate (102),

- filling the cavity to form the thick central part (103) of the membrane (100) by depositing a doped poly-silicon layer (135),

- patterning and etching the doped poly-silicon layer

- depositing an oxide insulating layer and patterning it to form a support to the membrane (100), and a first horizontal gap between the membrane thin peripheral part (108) and substrate (102)

- polishing the oxide insulating layer and the poly-silicon layer

- bonding an SOI wafer having a device layer (142) on it to the substrate (102),

- drilling a hole at the rear side of the substrate (102), by patterning and etching the rear side of the wafer until the thin dry oxide sacrificial layer is exposed,

- etching the dry oxide sacrificial layer (134) from the opened hole to release and enable the thick central part (103) of the membrane (100) to move freely,

- bonding a second wafer to the rear side of the substrate (102) in vacuum to form a vacuum sealed cavity at the rear side of the membrane (100), the cavity being a second horizontal gap

- providing electrical contacts to the substrate (102) and the membrane (100),

- wherein the substrate (102) and the membrane (100) are conductive, and

- wherein the side walls (1001) of the thick central part of the membrane (100) and vertical side walls (104) of the incised substrate (102) are configured for application of an actuation voltage.