Background technology
Ultrasonic transducer for medical imaging has the various characteristics for causing to produce high-quality diagnostic image.Including
Wide bandwidth and at ultrasonic frequencies for the high sensitivity of low-level acoustic signal.Traditionally, with these characteristics and
Thus the piezoelectric for ultrasonic transducer is made up of PZT and PVDF materials, wherein PZT is most preferred.However, ceramic
PZT material needs the manufacture for including significantly different and complicated cutting-up (dicing), matching layer bonding, filler, plating and interconnection
Technique and substantial amounts of process is needed, it is all these to cause less than expected transducer stackable unit yield.And, this system
Make the cost that complexity increased final transducer probe.With the main frame (mainframe) of ultrasonic system become it is less simultaneously
And it is main by field programmable gate array (FPGA) and for the software of majority signal processing function controlling, system host into
This is reduced with the size of system.Now, the ultrasonic system of cheap portable, desktop and hand-held form can be obtained.As a result,
The percentage that the cost of transducer probe accounts for system synthesis sheet is growing, and for the higher component count array of 3D imagings
Appearance accelerate this growth.It is therefore desirable to be able to the yield and low cost manufacture transducer array to improve, to promote
For the demand of inexpensive ultrasonic system.
Recent development already leads to that the prospect of medical supersonic transducer can be manufactured by semiconductor technology.It is desired
It is that these techniques should be identical with the technique for being used to manufacture the circuit needed for ultrasonic probe, such as CMOS technology.These development are
Jing generates micro-machined ultrasonic transducer or MUT.MUT is manufactured according to two kinds of designs, one kind is using with piezoelectricity
The semiconductor layer (PMUT) of characteristic and another kind is using the diaphragm with battery lead plate and substrate for showing capacity effect
(CMUT).CMUT transducers are the small chaffy devices with electrode, and it is by the acoustical vibration of the ultrasonic signal for being received
Be converted to modulating capacitor.In order to transmit, the capacitance charge to being applied to electrode be modulated the diaphragm to vibrate the device and
So as to transmit sound wave.Because these devices are manufactured by semiconductor technology, therefore the device generally has 10-200 micrometer ranges
Yardstick, but can be the device diameters for reaching 300-500 microns.Many so single CMUT can be connected together simultaneously
And integrally operate as single transducer element.For example, four to ten six single CMUT can be coupled together with whole
Body ground is used as single transducer element.Typical 2D transducer arrays at present will be with 2000-3000 piezoelectric transducer element.
When CMUT arrays are fabricated to, 1,000,000 CMUT units will be used more than., it is surprising that earlier result indicate that, this
The yield of (fab) CMUT arrays of the semiconductor manufacturing of size should be than the PZT array that formed by thousand of element of transducers
Yield is significantly improved.
CMUT is initially designed to and is operated according to now referred to as " not subsided (uncollapsed) " pattern.With reference to Fig. 1,
Typical not subsided CMUT transducer unit 10 is illustrated with section.On the substrate 12 of CMUT transducer units 10 and such as silicon
Multiple similar adjacent cells are manufactured together.Diaphragm or film 14 can be supported on by insulation support body 16 by made by silicon nitride
Substrate top, the insulation support body 16 can be made up of silica or silicon nitride.Chamber 18 between film and substrate can be
What air or gas were filled, or evacuate wholly or in part.Such as golden conductive film or layer 20 is formed on the diaphragm
Electrode, and similar film or layer 22 form electrode on substrate.The two the electrode shapes separated by dielectric chamber 18
Into electric capacity.Acoustic signal make film 14 vibrate when, the change of electric capacity can be detected, so as to by sound wave transducing be corresponding telecommunications
Number.Conversely, the AC signal for being applied to electrode 20,22 will modulate the electric capacity, the film movement is made and so as to transmit acoustic signal.
CMUT units are substantially secondary (quadratic) devices, therefore acoustic signal typically applies the humorous of signal
Ripple, i.e. acoustic signal will be the twice of applied signal frequency.In order to prevent this secondary behavior, apply to two electrodes
Plus bias voltage, diaphragm is attracted to substrate by produced Coulomb force.This schematically shows in fig. 2, wherein DC biased electricals
Pressure VBIt is applied to bias terminal 24 and is connected to by applying the path of the high impedance Z of such as inductive impedance to AC signal
Membrane electrode 20.AC signal is capacitively coupled to membrane electrode or from the membrane electrode Capacitance Coupled from signal terminal 26.On membrane electrode
Positive charge (+) make film expansion when it is attracted towards the negative electrical charge on substrate 12.CMUT units are in this bias state
Secondary behavior is only weakly shown during lower continuous operation.
It has been found that in film expansion so that the plate of two reverse chargeds of the capacitive device is as close possible to one
When rising, CMUT is most sensitive.The close proximity of two plates will be produced the bigger coupling between acoustics and electrical signal energy by CMUT.Cause
And expect to increase bias voltage VBUntil between film 14 and substrate 12 dielectric interval 32 it is as little as possible to be able to maintain that operation
Under signal conditioning.In the embodiment for being constituted, should be at intervals of one micron or less magnitude.If however, applied
Bias voltage is too big, and because Van der Waals (Vander Wals) power is pasted together two plates of device, then the film can contact lining
Bottom, makes shorted devices.This stickup can occur when CMUT units are easily overdriven, and due to manufacturing tolerance change, for phase
Same bias voltage VBAlso can change between device.Although by being embedded in device electrode in electric insulation layer (such as silicon nitride)
Permanent stickup can be reduced, but when attempting in peak response range of operation not subsided CMUT, subsided and not subsided shape
The non-linear of operation is intrinsic defect between state.
Even if when the film is spaced by the dielectric biased to produce very little sub-micron, the sensitivity of CMUT also can be less than
Expected sensitivity.This is due to the fact that, although the electric charge at the center 32 of film is relatively close to opposite charges and by phase
For the opposite charges is significantly moved, but the electric charge in the periphery 34 that the film is supported by supporter 16 will be moved hardly
And thus almost it is not involved in the signal transducing carried out by device.A kind of method for eliminating the difference is to use not extending to support
The less membrane electrode 20 of body 16.This by the charge-limited on membrane electrode will participate in strongly film movement and thus by the device
The device center of the transducing that part is carried out.It is still necessary to that there is one or more electric conductor to apply bias voltage to membrane electrode 20
VBAnd AC signal is coupled into electrode or coupling carrys out the AC signal of self-electrode.These electric conductors must be very thin, tool
Oriented AC signal applies the yardstick of undesirable big impedance, so as to limit the sensitivity of device.
Because CMUT and other MUT are manufactured by semiconductor technology, they dimensionally with traditional ceramic PZT transducers
Element is compared inevitable little.As previously mentioned, the superficial dimension of MUT is with micron measurement.When compact transducer array is needed,
Such as it is used for ultrasonic blood vessel inner catheter and the cardiac transducers probe scanned between rib, small size is favourable.However, depositing
In the other application of such as abdominal applications, wherein expecting larger aperture and deeper penetrating.In such applications, larger chi
Very little array is preferred.A kind of method that the array of larger aperture is constructed using CMUT is that multiple CMUT components are abreast mutual
With or tiling to together with forming a big array.Although whole component can on a single substrate be formed, due to array
Size increase and there is a possibility that increase there is manufacturing defect.In order to avoid the generation of problems, preferably manufacture
The less sub-component that can be examined, then forms whole component by multiple flawless sub-components.Each sub-component is flat
Spreading piece (tile) includes the one-dimensional or two-dimentional CMUT arrays for manufacturing on one substrate.Then single substrate tiling part is right side by side
Together with the big array of formation.In the sub-component for aliging multiple arrays, it may be desirable to keep on whole component each element pitch,
Or interval, rather than adjoin another edge in a substrate and there is gap.This causes aperture on the surface of whole component
On can equably transition, just as linear array transducer on the assembly aperture stepping or it is towed step on into when institute it is real
Existing.Uniform pitch will avoid being produced when the gap at the part edge that tiles otherwise will move into active transition aperture artifact and not
Desired secondary lobe.
Specific embodiment
With reference to Fig. 3, the schematic sectional view of CMUT elements or unit 5 is depicted.CMUT units 5 include substrate layer 12, electricity
Pole 22, film layer 14 and membrane electrode ring 28, in figure 6 CMUT units of visible circular form.In this example, electrode 22 is configured
For circular and be embedded in substrate layer 12.In addition, film layer 14 fix relative to the top surface of substrate layer 12 and be configured/
Really it is dimensioned so as to limit spherical or cylindrical chamber 18 between film layer 14 and substrate layer 12.As previously mentioned, the list
Unit and its chamber 18 can limit substituting geometry.For example, chamber 18 can limit rectangle and/or square sectional, six
Side tee section, oval cross section or irregular section.
Bottom electrode 22 insulate typically with extra play (not shown) on its surface towards chamber.Preferred insulation
Layer is formed in underlayer electrode top and oxidenitride oxide (ONO) dielectric layer below membrane electrode.ONO dielectrics
Layer advantageously reduces charge accumulation on electrode, and the charge accumulation causes the unstability of device and acoustics output pressure
Drift and reduction.Entitled " the capacitance type micromachined ultrasonic transducer of the submission on the 16th of September in 2008 is equal in Klootwijk
The european patent application No.08305553.3 of (Capacitive micromachined ultrasound transducer) "
In be discussed in detail ONO dielectric layer manufactured on CMUT.The pre- of impact is kept to collapse for electric charge is more vulnerable to than not subsided device
Sunken CMUT, expects to use ONO dielectric layer.Disclosed part can be manufactured by CMOS compatible materials, such as Al, Ti, nitride
(such as silicon nitride), oxide (various grades), tetraethyl orthosilicate (TEOS), polysilicon and the like.In CMOS manufactures
In, for example oxide and nitride layer can be formed by chemical vapor deposition and metallization (electricity is arranged by sputtering technology
Pole) layer.Suitable CMOS technology is LPCVD and PECVD, and the latter has the relatively low operation temperature less than 400 DEG C.
Example technique for manufacturing disclosed chamber 18 is included in before the top surface of addition film layer 14 in film
Chamber is limited in the initial part of layer 14.Other manufacture details can be looked in United States Patent (USP) No.6,328,697 (Fraser)
Arrive.In the exemplary embodiment shown in Fig. 3, the diameter of the battery lead plate 22 with diameter greater than circular configuration of cylindrical chamber 18.
Electrode retaining collar 28 can have the identical external diameter of battery lead plate 22 with circular configuration, although being not required for such uniformity.Thus,
In an exemplary embodiment of the present invention, electrode retaining collar 28 is fixed relative to the top surface of film layer 14, with the battery lead plate with lower section
22 alignments.
Fig. 4 illustrates the CMUT units of the Fig. 5 when pre- collapsed mode is biased to, and wherein film 14 connects with the bottom surface of chamber 18
Touch.This is realized by applying DC bias voltages to two electrodes, such as by the voltage V for being applied to electrode retaining collar 28BAnd be applied to
Underlayer electrode 22 reference potential () it is indicated.Although the continuous plate-like in the no hole centered on also being formed of electrode retaining collar 28,
But Fig. 4 shows the reason for this is not required.When as shown in the figure film 14 being biased into its pre- collapsed mode, the center of film with
The bottom surface contact of chamber 18.Thus, the center of film 14 is not moved during the operation of CMUT.But film 14 positioned at chamber 18
The open space top of residue and the movement of the outer peripheral areas below ring electrode.It is described by the way that membrane electrode 28 is formed as into ring
The upper plate electric charge of the electric capacity of device is located at the area that motion and capacitance variations are shown when CMUT is as transducer manipulation of CMUT
Domain top.Thus, improve the coefficient of coup of CMUT transducers.
By applying the necessary bias voltage typically in 50-100 volt ranges, film 14 can be brought into and chamber 18
Use 36 mark bottom surface contact pre- collapsed mode.As voltage increases, using C meter the electric capacity of CMUT units is monitored.Electricity
The mutation of appearance shows that the film has collapsed to the bottom surface of chamber.The film can be with biased downward until it just touches the use 36 of chamber
The bottom surface of mark, or can further downward be biased to subsiding for the increase beyond minimal-contact.
It is to apply pressure to the top of film by the another way that film 14 is brought into its pre- collapsed mode.When chamber forming part
When vacuum or perfect vacuum, it has been found that the atmospheric pressure for applying 1 bar (bar) be enough to make film 14 subside in advance and with chamber 18
Bottom surface contacts.Also film 14 can be controllably made to subside in advance using the combination of pressure differential and bias voltage, this is for high
Air Collapse Pressure (such as 10 bars) it is effective compared with gadget.Another technology for making film 14 bias to its pre- collapsed mode is
The structure that film is physically held in its pre- collapsed mode is arranged or formed above film.In the preferred reality for ultrasonic transducer
In applying example, the structure forms the lens of transducer.
Semiconductor device substantial amounts of manufacture generally on chip.After fabrication, all parts must separate or cut into list
Body, so as to allow them to be respectively used in different components and product.The typical process of each part is monolithically cut out from chip
It is to use cutting-up.During cutting-up, it is contemplated that the chip of certain level.In order that the propagation in the crack of cutting-up generation is minimum
Change, " crack arrest " groove 40 as shown in Figure 5 is usually used.Shallow crack arrest groove 40 is just formed in the expected side of the part after cutting-up
The inner side of edge, as shown in the drawing.Groove 40 is used to prevent crack to propagate towards CMUT units 5 during cutting-up process, and from
Crack arrest groove to neighbouring active device and slicing edge have acceptable distance requirement.This causes slicing edge and active device
(CMUT) sizable distance for generally compared with the distance between each device, this is for being tiled and kept flat
It is undesirable for pitch continuity between spreading piece.The typical sizes (diameter) of CMUT units can be changed to from 30 microns
300 microns even more big.The size of selected CMUT units is by depending on the expected operating frequency of given transducer.Generally, will
In keeping the average distance between each device less than 5 to 10 microns or even below 3 microns to maximize array
The density of CMUT unit components.Except the top view at tiling part edge illustrated at the top of Fig. 5, also show in the bottom of the accompanying drawing
The side view of section, including substrate 12, the ASIC layers comprising micro-beam-former circuit and top CMUT layers are gone out.
Principle of the invention, in order to solve the problem, is etched using singulation (singulation).Due to adopting
Etch process does not have a chip, thus does not need crack arrest groove, it is not required that from crack arrest groove to active device and the edge of etching must
Need distance.The further result of the technique is that the slicing edge of etching can be very close to CMUT units (generally in single device
At a half-distance between part 5), this makes it possible to constant (continuous) pitch for keeping being cut into slices to another from a section.
Fig. 6 is the top view of CMUT arrays, and wherein dotted line 48 indicates the tortuous etched path for edge, and it is wrapped near CMUT units 5
Around (substantially along the shape of CMUT units).Adjacent tiling part is formed with the tortuous etched path of complementation so that the two are put down
Spreading piece will be combined together to another tiling part from a tiling part in rows of unit with constant pitch.
Because etching needs from section top vertically to extend completely through the bottom to substrate, it is preferably
Using deep reactive ion etch technique, it can form highly vertical edge on depth is extended in a controlled manner.It is a kind of
For do so technology in Fig. 7 a) to Fig. 7 d) and in illustrate.In this example, such as Fig. 7 a) shown in optional mask 50 be arranged at
In section, and the opening that is guided through in mask of plasma (sulfur fluoride) stream of high anisotropy and ion stream start
It is etched through section and its substrate 12.Plasma be partially etched into section in when, such as Fig. 7 b) as shown in
Deposit passivation layer 52 on groove.Passivation layer can be suitable polymer, such as polyimides or benzocyclobutane alkenyl polymer
(BCB).Passivation layer protects the wall of groove to prevent from being further exposed to chemical etchant.Anisotropic etchant will attack groove
Passivation layer at bottom, is sputtered away, and then, groove can be etched to deeper depth, such as Fig. 7 c) shown in.It is alternative
Ground, when groove is formed from top section (Fig. 7 b)), by chip upset and from section back side etch groove, such as
Fig. 7 d) shown in, and stop at polymer 52.Finally, polymer is sputtered away by etching.From top section landform
It is the ditch by making chip thinning from dorsal part (bottom) grinding and being partly formed up to arrival into the another alternative after groove
Groove.
Fig. 8 shows chip 60, and two CMUT tiling parts 62 and 64 are according to the present invention by etching by from 60 points of chip
It is cut into monomer.In this example, interconnection electrode 70 is formed on the top surface that each is cut into slices along the edge that each is cut into slices.Due to
Etch process can be controlled to form the edge of almost arbitrary shape or profile, thus the edge can closely along such as should
The shape of the CMUT units 5 in example.This enables the tiling part to be combined together along different orientation, such as interconnects electricity
On the phase homonymy as shown on chip 60 of pole, or tiling part alternately overturns the phase for causing interconnection electrode shown on the right side of such as accompanying drawing
On anti-both sides.The electricity for alternately allowing the interconnection of such as flexible circuit to access and be attached on multiple sides of assembled CMUT arrays
Pole 70.This is illustrated in the CMUT array components for completing of Fig. 9, wherein four tiling parts 62,64,66 and 68 are engaged in together
To form large aperture CMUT arrays, the pitch of wherein CMUT units is maintained uniformly on array.
Figure 10 show in block form the large aperture CMUT array transducings with tiling of constructed in accordance with the principles
The ultrasonic diagnosis imaging system of device probe 100 '.In this example, large aperture array 100 is formed by 16 tiling parts.Transducing
Device array 100 ' is one-dimensional or two-dimensional transducer element arrays, and it can sweep in three-dimensional in 2D planes or for 3D imagings
Retouch.Transducer array is attached to the micro-beam-former 112 in probe, and it is by CMUT array elements come the transmission of control signal
And reception.The micro-beam-former can at least to the signal by element of transducer group or " tile (patches) " reception
Carry out part beam forming, such as United States Patent (USP) 5,997,479 (Savord etc.), 6,013,032 (Savord) and 6,623,
Described in 432 (Powers etc.).The micro-beam-former is attached to transmitting/reception (T/R) switch by cable of popping one's head in
116, the switch switches between transmit and receive, and is not used in micro-beam-former and the transducer array is by leading
System beam-shaper protects main system beam-shaper 120 from the harm of high-energy transmission signal when directly operating.Micro-
Under the control of beam-shaper 112, from CMUT transducer arrays 100 ultrasonic beam transmission by be attached to T/R switch and
The transducer controller 118 of main system beam-shaper 120 is guided, and the transducer controller is received from user interface or control
The input of the user operation of panel processed 38.The One function controlled by transducer controller is the direction that wave beam is directed.Ripple
Shu Kecong (being orthogonal to) transducer array is directly guided forward, or is guided at different angles for broader visual field.Transducer
Controller 118 also controls to apply to the DC of CMUT units to bias, and unit membrane 14 is biased to pre- collapsed mode by the DC biass.
The signal of the part beam forming generated after micro-beam-former 112 is received is attached to main beam former
120, here is combined into the letter of complete beam forming from the signal of the part beam forming of the independent tile of element of transducer
Number.For example, main beam former 120 can have 128 passages, and each passage is beaten or hundreds of CMUT transducer units shapes from several
Into tile receiving portion beam forming signal.So, received by the thousands of element of transducers of CMUT transducer arrays
Signal can effectively facilitate the signal of single beam forming.
The signal of the beam forming is attached to signal processor 122.Signal processor 122 can be located in a variety of ways
The received echo signal of reason, such as bandpass filtering, extraction, I and Q component are separated and harmonic signal is separated, the harmonic wave letter
Number separate for segregated linear signal and nonlinear properties, enabling the nonlinear echo letter that identification is returned from tissue and microvesicle
Number.Signal processor can also carry out additional signal enhancing, and such as spot suppresses, signal synthesis and noise are eliminated.The letter
Bandpass filter in number processor can be tracking filter as above, and wherein its passband is with from increased depth
Reception of echoes signal and from high frequency band to lower band slide, and then refuse from bigger depth higher frequency noise,
Wherein these frequencies do not have anatomic information.
Handled signal is attached to B-mode processor 126 and doppler processor 128.B-mode processor 126 is sharp
With amplitude detection for such as intracorporeal organ organize or blood vessel internal structure imaging.The B-mode image of body structure can
Formed according to the combination of harmonic mode or basic model or both, such as United States Patent (USP) 6,283,919 (Roundhill etc.) and U.S.
Described in state's patent 6,458,083 (Jago etc.).The process of doppler processor 128 comes self-organization and blood flow
Instantaneous distinguishing signal, for detecting the motion of matter of such as haemocyte flowing in picture field.Doppler processor is generally wrapped
Wall filter is included, its parameter is provided through and/or refuses the echo returned from the material of internal selection type.For example,
Wall filter can be configured with pass-band performance, and the characteristic leads to the signal of the relative short arc of the material from fair speed
Cross, but refuse the relatively strong signal from relatively low or zero velocity material.This pass-band performance will be made from fluid flow blood
Signal passes through, but refuses the signal of the object of the neighbouring static or slow motion from such as heart wall.Contrary characteristic will make
Signal from the motor tissue of heart passes through, but refuses the signal of blood flow, and this is referred to as tissue doppler imaging, detection
With the motion for describing tissue.Doppler processor receives and processes the instantaneous discrete echo signal sequence of the difference in picture field
Row, from the echo sequences of specified point, it is called ensemble (ensemble).Quickly connect in relatively short time interval
The echo signal group energy that continued access is received is used for the Doppler shift frequency of assessment of flow blood, and wherein Doppler frequency is corresponding with speed
Relation indicates blood flowing speed.The echo signal group received in longer period of time be used to estimate the blood or slow that flowing is slower
The speed of the tissue of slow motion.
The structure and motor message generated by B-mode and doppler processor is attached to scan converter 132 and Duo Ping
Face reformatting device 144.Spatially relation arranges echo signal to the scan converter, and the echo signal is closed according to the space
System is received with expected picture format.For example, scan converter can by echo signal be arranged to two-dimentional (2D) sector formats or
Cone three-dimensional (3D) image.Scan converter by B-mode structural images overlay color, the color corresponding to each point in picture field at
The motion being consistent with their Doppler evaluation speed, so as to generate the colour of the tissue and blood flow movement depicted in picture field
Doppler image.Many planar reformat devices turn the echo received at each point of the common plane from the volumetric region of body
The ultrasonoscopy of the plane, such as United States Patent (USP) 6 are changed to, described in 443,896 (Detmer).Volume rendering device 142 is by 3D
The echo signal of data set is converted to the projection 3D rendering seen from given reference point, such as United States Patent (USP) 6,530,885
Described in (Entrekin etc.).2D or 3D rendering are painted from scan converter 32, many planar reformat devices 44 and volume
Device processed 142 is attached to image processor 130 to further enhance, buffer and temporarily store, on image display 40
Show.In addition to for imaging, the Hemodynamic environment angle value generated by doppler processor 128 is attached to fluid quantitative processor
134.The fluid quantitative processor generates measuring for different flow regimes, the volume fraction of such as blood flow.The fluid quantitative processor
Can be from the receives input of user's control panel 38, by the point in the anatomical structure for measuring in such as image.From fluid quantitative
The output data of processor is attached to graphic process unit 136 for reproducing together with image measurement number on the display 40
Value.It is stacked for showing together with ultrasonoscopy that graphic process unit 136 can also generate figure.These figures are stacked can be included
Standard identification information, patient's name, the date and time of image, imaging parameters etc..For such purposes, graphic process unit
From the receives input of user interface 38, the patient's name for such as keying in.The user interface is also coupled to mission controller 18 to control
From transducer array 100 and therefore generated by transducer array and ultrasonic system image ultrasonic signal generation.The use
Family interface is also coupled to many planar reformat devices 144 for selecting and controlling (MPR) image of many planar reformats
Show, this can be used to perform the quantitative measurment in the picture field of MPR images.