sensors Article Multielement Ring Array Based on Minute Size PMUTs for High Acoustic Pressure and Tunable Focus Depth Eyglis Ledesma , Iván Zamora , Arantxa Uranga and Núria Barniol * Departament d’Enginyeria Electrònica, Universitat Autónoma de Barcelona, 08193 Bellaterra, Spain; eyglis.ledesma@uab.es (E.L.); ivan.zamora@uab.es (I.Z.); arantxa.uranga@uab.es (A.U.) * Correspondence: nuria.barniol@uab.es Abstract: This paper presents a multielement annular ring ultrasound transducer formed by indi- vidual high-frequency PMUTs (17.5 MHz in air and 8.7 MHz in liquid) intended for high-precision axial focalization and high-performance ultrasound imaging. The prototype has five independent multielement rings fabricated by a monolithic process over CMOS, allowing for a very compact and robust design. Crosstalk between rings is under 56 dB, which guarantees an efficient beam focusing on a range between 1.4 mm and 67 µm. The presented PMUT-on-CMOS annular array with an overall diameter down to 669 µm achieves an output pressure in liquid of 4.84 kPa/V/mm2 at 1.5 mm away from the array when the five channels are excited together, which is the largest reported for PMUTs. Pulse-echo experiments towards high-resolution imaging are demonstrated using the central ring as a receiver. With an equivalent diameter of 149 µm, this central ring provides high receiving sensitivity, 441.6 nV/Pa, higher than that of commercial hydrophones with equivalent size. A 1D ultrasound image using two channels is demonstrated, with maximum received signals of  7 mVpp when a nonintegrated amplifier is used, demonstrating the ultrasound imaging capabilities.  Citation: Ledesma, E.; Zamora, I.; Keywords: PMUTs; annular array; ring array; ultrasound; AlN; ultrasound imaging; pulse-echo; Uranga, A.; Barniol, N. Multielement PMUT-on-CMOS Ring Array Based on Minute Size PMUTs for High Acoustic Pressure and Tunable Focus Depth. Sensors 2021, 21, 4786. https://doi.org/ 1. Introduction 10.3390/s21144786 Ultrasound is widely used as an excellent noninvasive diagnostic tool for nondestruc- tive testing and medical imaging. Nowadays, small ultrasound probes are being extensively Academic Editor: Anthony N. Sinclair pursued in areas such as in-body controllable catheter-based imaging for intravascular imaging [1–4], specific heat treatments based on high-intensity focused ultrasound (HIFU Received: 22 June 2021 Accepted: 10 July 2021 systems) [5,6], or brain stimulation for in vivo experiments [7,8]. In addition to these appli- Published: 13 July 2021 cations, power-free implantable prostheses for sensing biological parameters in animals or humans are also a rapidly evolving field of research towards digital medicine and in view Publisher’s Note: MDPI stays neutral of elderly population growth. Recently, powering and data transfer for these devices by with regard to jurisdictional claims in ultrasound is being studied and presented as one of the alternatives in comparison with published maps and institutional affil- radiofrequency and inductive links [9–11]. iations. For all these cases, tiny ultrasound transducers with very controllable and tunable focus depth are required. Among the different ultrasound probes, annular rings provide by layout this capability. Their dynamic focusing along the axial direction and their symmetry produce an acoustic pattern with a high spatial resolution (lateral and axial) and high Copyright: © 2021 by the authors. levels of output pressure which have been efficiently used in most of the above-mentioned Licensee MDPI, Basel, Switzerland. applications [1,3–6,8,9]. This article is an open access article Annular rings for ultrasound systems have been fabricated with bulk piezoelectric distributed under the terms and materials in their thickness vibration mode. However, their complex fabrication process conditions of the Creative Commons would limit their use in advanced ultrasound systems which require minute sizes, high Attribution (CC BY) license (https:// reproducibility, cost-efficiency, and low power consumption [12,13]. Advances in the fabri- creativecommons.org/licenses/by/ cation of micromachined ultrasonic transducers using MEMS-based technologies, either 4.0/). capacitive (CMUTs) or piezoelectrical (PMUTs) with the capability of direct integration Sensors 2021, 21, 4786. https://doi.org/10.3390/s21144786 https://www.mdpi.com/journal/sensors Sensors 2021, 21, 4786 2 of 18 with CMOS, in a batch processing approach, decrease manufacturing cost and allow a reduction in size and increase in compactness of the overall ultrasound system [12,14,15]. Ring-shaped CMUT arrays have been widely described in the literature [3–5,12,14]. How- ever, their eligibility can be affected by the high power consumption needed in the CMUT transducers. On the contrary, monolithic ultrasound systems based on PMUTs open the way to power-efficient single-chip ultrasound systems [15,16]. In PMUTs, the ultrasound wave is produced due to the flexural movement of the membrane, and its operational frequency will depend on both the different thicknesses of the material and the size of the transducer, allowing in this way different operation frequencies using different device layouts but with the same technological approach. Few examples of continuous annular rings under this flexural approach have been reported [6]. However, continuous annular array fabrication could be a challenge due to its dimensional constraints. In order to avoid this and increase the achievable output pressure at the acoustic axis, a finite number of small ultrasound transducers configured in an annular ring can be used (see for example [1,17]). The main advantage in this multielement ring configuration is given by the capability to decouple the acoustic wave frequency, due exclusively to the single element, from the overall ring diameter, keeping a longer-range tunable focus depth, as is shown in Section 2. Taking these demands into consideration, we have designed a multiple concentric annular ring ultrasonic transducer composed of multiple high-frequency piezoelectrical micromachined ultrasound transducers (PMUTs) capable of being monolithically integrated with CMOS technology [16]. The fabricated multielement ring ultrasound array provides high accurate tunable focus depth, high acoustic output pressure, and minute size which is operable in the 10 MHz frequency range in a liquid environment and can be a candidate for the above-mentioned applications. In addition, we have characterized the central ring of the array as an ultrasonic receiver, showing higher receiver sensitivity with smaller spatial averaging effects than commercial hydrophones with similar size. This paper is organized into four sections: Section 2 explains the multielement ring array design and its benefit over the continuous ring array using analytical equations and FEM simulations (COMSOL Multiphysics); predictions of the acoustic performance with Field II software are also provided. Section 3 shows the experimental results, in- cluding the electrical and acoustic characterization and line-scan ultrasound imaging as a demonstration of the full transceiver ultrasound system. Finally, Section 4 concludes the paper. 2. Materials and Methods 2.1. Multielement Ring Array Design As already explained, the main advantage of an annular ring is its capability to be axially focused; focus depth tunability, beam diameter at the focus point, and range of achievable pressure are its main parameters. The focus depth, Fd , is defined by Equation (1), where S is the transducer area (whatever its shape), λ is the wavelength, f is the frequency, and c is the sound velocity in the propagation medium [18]. It is evident that transducers with the same area will produce a greater focus depth with higher frequency. S S× f Fd = = (1) 4λ 4×c For the design of an efficient annular ring, the first question is how we can achieve the highest frequency using the same transducer area, which will give us the greatest focus depth. In our approach, we must consider flexural resonators as we are using a MEMS-based approach with AlN as the piezoelectrical material. We will analyze this focus depth capability considering a continuous annular ring array in comparison with a multielement annular ring array. Figure 1 top inset shows a conceptual schematic of a single element from a continuous annular ring array (green) and a single element (squared PMUT) that will form a multielement ring array, (blue). Here 2D, 2d, and l are the outer Sensors 2021, 21, x FOR PEER REVIEW 3 of 18 Sensors 2021, 21, 4786 3 of 18 will form a multielement ring array, (blue). Here 2D, 2d, and l are the outer diameter, inner diameter, diameter, inner diameter, and width andring of each width in of theeach ring in annular the annular array; are thea and a and warray; side wand arediagonal the side and diagonal of each square PMUT in the multielement of each square PMUT in the multielement array. array. Theoreticalfrequency Figure 1. Theoretical frequencyratio, ratio, fs /fafrom fs/fa, , from two two single single elements elements of anofannular an annular arrayarray and aand mul-a tielement multiring array with the same area according to Equation (4). Inset: Only one multielement multiring array with the same area according to Equation (4). Inset: Only one annularannular ring and one ring and single square one single PMUT square fromfrom PMUT the annular the annular and and multielement ring ring multielement arrays are shown. arrays are shown. First, the resonance resonance frequency frequencyfor forthe thefirst firstflexural flexuralmodemodefor forthe theannular annularring array ring arrayis giving is byby giving Equation Equation (2),(2),where where λij_annular is a isdimensionless λij_annular a dimensionless parameter parameter associated withwith associated the the vibration vibration mode mode (i, j),(i,Drj),isD r isflexural the the flexural rigidity, rigidity, and µand µ is is the the per mass mass perarea unit unit[19]. areaIntui- [19]. Intuitively, tively, higher higher resonanceresonance frequencies frequencies could could be obtained be obtained with smaller with smaller devices,devices, but the but pa- the parameter rameter λij_annularλij_annular is dependent is dependent on the ratio on the ofratio d/D andof d/D and boundary boundary conditions, conditions, so in so in conse- consequence, quence, the same the same frequency frequency can be can be obtained obtained for rings for rings with with the samethe same widthwidth (no matter (no matter how how big the outer diameter is), even in the case big the outer diameter is), even in the case of flexural resonant rings. of flexural resonant rings. λij _ annular2 2 DD s λij_annular a = i =i = . . j =1,1,2,2, . . r r f a f= 1,1,2,2, .j = . (2) (2) 2πD2π D 2 2 µμ Second, for for aa multielement multielementring ringarray, array,the theresonance resonancefrequency frequencyisisfixed fixedbyby the indi- the in- vidual PMUTs, dividual PMUTs,defined definedby byEquation Equation(3)(3)for for the the first first flexural mode considering considering square square PMUTs [20]. PMUTs [20]. s 35.99 Dr f s = 35.99 D (3) f =2πa2 µ r s (3) 2π a 2 μ Using Equations (2) and (3), the ratio fs /fa for the same layer stack is given by Equation (4) and computed Using (2) andin(3), Figure 1 for different the ratio d/Dsame fs/fa for the ratioslayer considering stack isclamped given byboundaries Equation (4) where and the parameter 2 computed in Figure λij_annular 1 forhas been extracted different d/D ratios from Table 11-2clamped considering in [19]. The multielement boundaries wherering the array achieves parameter frequencies λij_annular 2 has beenhigher thanfrom extracted the annular Tables 1array and 2when the in [19]. ThePMUT side is much multielement ring smaller than D.frequencies array achieves Dotted lineshigher show thanthe case the when annularthearray ring when width theis equal PMUT a (minimum to side is much side of the smaller squared than D. Dotted PMUT) lines w, giving orshow for all the case d/D ratios when an width the ring improvement is equalintomultielement a (minimum multiring side of thefrequency squared PMUT) of 1.6×or orw, ×, respectively. 3.3giving for all d/DThis increase ratios in the resonance an improvement frequency in multielement is translated into a higher focal length. multiring frequency of 1.6× or 3.3×, respectively. This increase in the resonance frequency is translated into a higher focal length. 35.99 f s /f a = i = 1, 2, . . . j = 1, 2, . . . (4) ( a/D)2 × λij_annular 2 Sensors 2021, 21, x FOR PEER REVIEW 4 of 18 Sensors 2021, 21, 4786 4 of 18 fs 35.99 = i = 1, 2,... j = 1, 2,... fa ( D) ⋅ λ 2 (4) a 2 ij _ annular Figure 2a shows an optical image of the proposed multielement ring array ultrasound FigureThe transducer. 2a shows PMUTs anare optical image arranged inof the proposed irregular multielement polygons ring array that are connected ultrasound through the top electrode forming five concentric channels. The bottom electrode is commonthrough transducer. The PMUTs are arranged in irregular polygons that are connected for all the topdevices, PMUT electrode andforming the gapfive concentric between channels. consecutive The bottom elements electrode is common for is 25 µm. all PMUT devices, and the gap between consecutive elements is 25 µm. Figure2.2.(a) Figure (a)Optical Opticalimage imageofofthe themultielement multielementring ringarray arraytransducer transducerand andschematic schematicrepresentation representation ofofthe thecontinuous continuousrings ringsover overit;it;(b) (b)zoom zoomofofthe theindividual individual4040µm µmAlN AlNPMUT; PMUT;(c) (c)AA’ AA’cross-section cross-section of AlN-PMUT. of AlN-PMUT. Everysingle Every singleelement elementconsists consistsofofaasquared squaredAlN-PMUT AlN-PMUTwith withaa4040µm µmside, side,fabricated fabricated usingthe using theMEMS-on-CMOS MEMS-on-CMOS SilTerra SilTerra technology technology [16,21]. [16,21]. The The top top electrode electrode size size was wasopti- op- mized to timized tomaximize maximizethe themembrane membranevelocity velocityandandconsequently consequentlythe output the output pressure. pressure. A 1.3 A µmµm 1.3 AlNAlNpiezoelectric piezoelectricmaterial waswas material deposited depositedby by physical vapor physical deposition vapor (PVD) deposition and (PVD) sandwiched and sandwiched between betweentwo two Al electrodes (0.35(0.35 Al electrodes µm thickness top electrode µm thickness and 0.4 top electrode and µm 0.4thick- µm ness bottom electrode). A 1.5 µm Si N layer was thickness bottom electrode). A 1.5 µm Si3 N4 layer was deposited 3 4 deposited with a low-temperature a low-temperature plasma-enhancedchemical plasma-enhanced chemicalvapor vapordeposition deposition(PECVD) (PECVD)process; process;ititacts actsas asthe theelastic elasticlayer layer andseals and sealsthe thecavity. cavity.Figure Figure2c2cshows showsaacross-section cross-sectionprofile profile(AA’) (AA’)ofofthe thesquare squarePMUT PMUT device,and device, andTable Table 1 summarizes 1 summarizes thethe material material properties properties usedused in FEM in FEM COMSOLCOMSOL simula- simulations tions and and the the principal principal geometricgeometric dimensions. dimensions. Table1.1.Material Table Materialproperties andand properties thickness used in thickness FEM used insimulations. FEM simulations. Properties Properties Geometric Geometric PMUT Layer PMUT Layer Mat. Young’s Modulus (GPa) Density 3 Side (µm) Mat. Young’s modulus (GPa) Density (kg/m ) Side (µm) Thick. Thick.(µm) (µm) Substrate SiO2 70 (kg/m 2200 3) 60 2 Substrate Bottom Elec. SiO Al 2 70 70 2200 2700 60 46 0.42 Piezoelectric AlN 1 279 3230 60 1.3 Bottom Elec. Al Top Elec. Al 70 70 2700 2700 46 28.3 0.4 0.35 Piezoelectric Passive AlN Si3 N41 279 250 3230 3100 60 60 1.3 1.5 Top 1 The Elec. Al 70 2700 2 28.32 piezoelectric coefficients e33 and e31 used in COMSOL are 1.55 C/m and −0.6 C/m respectively.0.35 Passive Si3N4 250 3100 60 1.5 1 The piezoelectric coefficients e33 and e31 used in COMSOL are 1.55 C/m2 and −0.6 C/m2 respec- The first mode shape and its resonance frequency for a square 40 µm AlN PMUT tively. were obtained in COMSOL Multiphysics; the value of 27.7 MHz was given, which is close to the value computed using Equation (3), 27.8 MHz. Dynamic simulations in a liquid environment (Fluorinert, FC-70, with a density ρ = 1940 kg/m3 and the sound velocity c = 685 m/s) give a maximum displacement at 13 MHz. Sensors 2021, 21, 4786 5 of 18 In order to compare the performance of this multielement ring array transducer with an equivalent continuous concentric ring, we considered the circles that surround the single-PMUT elements. The shape of each ring is close to an irregular polygon, and it is not possible to obtain rings that include all PMUTs of the polygon exactly. Figure 2a shows in different colors the geometric representation corresponding to each continuous ring, where D and d are the outer and the inner radii, respectively, and w is the width. The values of D and d were computed trying to include the highest number of PMUTs of each irregular polygon into the continuous ring. The width, w, corresponds to the PMUT diagonal (the fur- √ thest PMUT point inscribed in the continuous ring) and is 56.6 µm ( (40 µm)2 + (40 µm)2 ) (see Figure 2b). A gap of 8.4 µm between rings is obtained. Decreasing the ring width to w = a, higher resonance frequency for the continuous ring can be achieved at the expense of not including the largest number of PMUTs within it. Table 2 summarizes the computed dimensions taking into account all these considerations. Table 2. Focus depth comparing multielement ring array with an equivalent continuous ring array using Equation (1) and the resonance frequencies obtained from the COMSOL simulations. The Fd is calculated with all actuated smaller rings. (c = 685 m/s). Dimensions Focal Point → Fd (µm) Number of Rings Continuous Ring; Multielement Ring; D (µm) d (µm) f = 2.3 MHz f = 11.3 MHz 1 74.28 17.68 13.7 67.4 2 139.28 82.68 50.3 247.3 3 204.28 147.68 109.2 536.6 4 269.28 212.68 190.4 935.4 5 334.28 277.68 293.8 1444 As the parameter λij in Equation (2) and the added virtual mass are only known for defined d/D ratios [19,22], we performed some FEM simulations with COMSOL to find the resonance frequencies for the first continuous ring in air and liquid, obtaining 8.9 and 2.3 MHz respectively. The resonance frequency for the continuous ring is 3 times lower than that of the multielement ring (27.8/8.9), which is close to the fsquared /fannular ratio when the annular width corresponds to the square PMUT diagonal (see Figure 1 when l = w). Figure 3 shows the frequency response in liquid for the first ring corresponding to the continuous ring and the multielement ring array. The normalized pressure maps for both rings, considering a propagation medium of 200 µm radius, are shown in Figure 3b,c. The multielement array achieves a higher focus depth in comparison to a continuous ring (considering the same actuated rings) due to its higher operation frequency (see Fd in Table 2), concentrating the acoustic pressure in a narrower beam, but it is affected by side lobes. The focus depth (Fd ) is presented in Table 2 for both the multielement ring array device and the continuous concentric ring device considering the same area in both cases. The advantage of focus depth control for the multielement ring array from 67 µm to 1.44 mm is clearly demonstrated. Sensors2021, Sensors 2021,21, 21,xxFOR FORPEER PEERREVIEW REVIEW 66 ofof 18 18 Sensors 2021, 21, 4786 6 of 18 (b) (b) (a) (a) (c) (c) Figure3. Figure Figure 3.FEM 3. FEM COMSOL FEMCOMSOL COMSOLdynamicdynamic simulation dynamicsimulation simulation inininliquid liquid for for liquid the the for the first first continuous continuous first continuous ring ring and and ring thethe the and first first multielement multielement first ring: ring: multielement (a) (a) ring: frequency frequency (a) frequencyresponse response responsewhere where where left-bottom left-bottom left-bottom (red (red curve) curve) (red curve) corresponds corresponds corresponds tototocontinuous continuous ring continuousring and ringand right-top andright-top (bluecurve) right-top(blue curve)corresponds corresponds corresponds totomultielement to multielementring; multielement ring;(b) ring; (b)continuous (b) continuousring continuous ringpressure ring pressuremap pressure mapat map atat2.3 2.3MHz; 2.3 MHz;(c) MHz; (c)multielement (c) multielementring multielement ringpressure ring pressuremap pressure mapat map atat11.3 11.3MHz. 11.3 MHz. MHz. 2.2. 2.2.Acoustic AcousticPerformance PerformanceSimulation Simulationfor forthe thePMUT PMUTArray Array Acoustic Acousticsimulations simulationsusing usingField [23,24]were FieldIIII [23,24] [23,24] wereperformed were performedto performed topredict to predictthe predict theacoustic the acoustic acoustic performanceof performance performance ofthethemultielement multielementring ringarray array working arrayworking workingin in liquid inliquid (FC-70, (FC-70,ccc===685 liquid(FC-70, 685 m/s) 685m/s)m/s) atatat aa acenter center center frequency frequency frequency of of of11.3 11.3 11.3 MHz MHz MHz (according (according (according to to FEM tothe the the FEM FEM COMSOL COMSOL COMSOL simulations simulations simulations shown shown shown inFig- in in Fig- Figure3). 3). ure3). ure Figure44shows Figure showsthe shows the the 2D2D 2D normalized normalized normalized pressure pressure pressure map map map from from from 20 to 20µm 20 µm µm to to33mmmm 3 along mm along along theaxial the the axial axial direction directionand direction andfrom and from−500from −500µm − 500 µmto µm to500 500µm to 500 µm µmlaterally, laterally, laterally,keeping keepingthe keeping thearray the array arraycenter center centeratat(0, at (0, (0,0)0)coordi- coordi- 0) coordinates. nates.Note nates. Note Notethat, that,as that, asTable as Table Table22shows, 2 shows,byshows, by byplaying playing playingwith withthe with thenumber the number numberof of ofactive active activerings, rings, rings,the thefocalthe focal focal point point pointcancanbe can be be changed changed changed without without without any delay. anyextra any extra extra delay.delay. Thebeam The The beam beam focusing focusing focusing rangerange range from fromthese from these theseField Field IIField II simulations IIsimulations simulations isisaabitishigher, bit a bit higher, higher, reaching reaching reaching 1.6mm, 1.6 1.6 due mm, mm, dueto due to the the tosimulation the simulation simulation considering considering considering the thereal the real real elementelement element distribution distribution distribution inside inside inside thethe the multielement multielement multielement ring ring ring array array array (gap (gap (gap spacesbetween spaces spaces betweenelements), between elements), elements), which whichisis which small issmall smallandand andisis not isnot considered notconsidered consideredin in Equation inEquation Equation(1). (1). (1). Figure Figure4. 4. Field FieldIIIIIIsimulated 4.Field simulated pressure simulatedpressure pressuremap map mapforfor the forthe multielement themultielement multielementarray array when arraywhen transmitting whentransmitting with transmittingwith (a) with(a) all (a)all rings, allrings, (b) rings,(b) four (b)four rings fourrings rings Figure (ring #1 (ring#1 (ring to ring #1totoring #4), ring#4), (c) #4),(c) three (c)three rings threerings (ring rings(ring #1 (ring#1to ring #1totoring#3), ring#3),(d) #3),(d)two (d)tworings tworings(ring #1 rings(ring to (ring#1 ring #2), #1totoring and ring#2), (e) #2),and one and(e) ring (e)one (ring onering #1). ring(ring Axis: (ring#1). Scan #1).Axis: Axis: in x-direction Scan Scan fromfrom ininx-direction x-direction − 0.5 -0.5 from to 0.5 -0.5totomm, z-direction 0.5mm, 0.5 mm, fromfrom z-direction z-direction 0.020.02 from to 3 to 0.02mm. to33mm. mm. On the Onthe On other theother hand, otherhand, controlling hand,controlling the controllingthe applied theapplied signal appliedsignal phase, signalphase, the phase,the focal thefocal point focalpoint can pointcan also canalso alsobebe be modified (electronic modified(electronic modified focusing), (electronicfocusing), achieving focusing),achieving higher achievinghigher pressure higherpressure levels pressurelevels than levelsthan at the thanatatthe natural thenatural focus. naturalfocus. focus. In Figure InFigure In 5a, Figure5a, the 5a,the red thered axis redaxis shows axisshows the showsthe dependence thedependence dependenceof of the ofthe transmission thetransmission improvement transmissionimprovement improvementon on the onthe the acoustic focusing factor (Sac), defined as Sac = Fac/N0 where Fac is the actual focus and Sensors 2021, 21, x FOR PEER REVIEW 7 of 18 Sensors 2021, 21, 4786 7 of 18 acoustic focusing factor (Sac), defined as Sac = Fac/N0 where Fac is the actual focus and N0 is the N0 natural focus (1.6 is the natural mm). focus (1.6 In the simulations, mm). the delays In the simulations, were applied the delays to all the were applied to ele- all the ments that make elements that up makeeach upring. eachAsring. expected, when the As expected, focusthe when point focusis close pointtoisnatural close tofocus, natural there is nothere focus, transmission sensitivitysensitivity is no transmission enhancement (value close enhancement to 1),close (value so totoensure 1), so attoleast ensure twice as much at least twice theastransmission pressure, the much the transmission acousticthe pressure, focusing acoustic factor should focusing be 0.8 factor (and be should 0.8 (and consequently consequently the focal pointthewill focal be point willFurthermore, 1.3 mm). be 1.3 mm). whenFurthermore, when it is focused, theit acoustic is focused, the acoustic energy energy is is concentrated in concentrated narrow beams, in decreasing narrow beams, decreasing the focus width, the andfocus width, and consequently consequently improving improving the capability the capability to detect to detect small targets. On the small othertargets. hand, inOn the other Figure 5a, thehand, blue in axisFigure shows5a, thethe blue axis shows dependence of thethe dependence beamwidth of dB at −6 theonbeamwidth the acoustic at − 6 dB on factor. focusing the acoustic In thisfocusing case, thefactor. In this is beamwidth case, the when wider beamwidth is wider the acoustic when the focusing acoustic factor focusing is close to 1, as factor ex- is close to 1, as expected. pected. (a) (b) Figure Figure 5. Field 5. Field II simulations: II simulations: (a) dependence (a) dependence of the of the transmission transmission sensitivity sensitivity improving improving factor factor (red(red curve) and the beamwidth (blue curve) on the acoustic focusing factor; (b) cross-section curve) and the beamwidth (blue curve) on the acoustic focusing factor; (b) cross-section in lateral in lateral direction direction at two at two different different focal focal points points (500 (500 and and 700 700 µm, correspondingtotoacoustic µm,corresponding acousticfocusing focusingfactors factors of of 0.3 0.3and and0.44, 0.44,respectively). respectively). Sensors 2021, 21, x FOR PEER REVIEW 8 of 18 Sensors 2021, 21, 4786 8 of 18 Figure 5b shows the acoustic pressure field profile along the lateral direction at dif- ferent focus points, including the natural focus (N0). As can be seen, the multielement ring array Figure 5b shows is affected the acousticofpressure by the generation unwantedfield profile lobes (sidealong lobes).the Forlateral focus direction at dif- depths greater ferent focus points, including the natural focus (N ). As can be seen, the than 500 µm, these side lobes are below −15 dB, which is proven to be the required dy- 0 multielement ring array namic rangeisfor affected by [25]. imaging the generation of unwanted lobes (side lobes). For focus depths greater than 500 µm, these side lobes are below −15 dB, which is proven to be the required 3.dynamic range for Experimental imaging Results and[25]. Discussion 3.1. Electrical Characterization 3. Experimental Results and Discussion 3.1. The electrical Electrical characterization in the air was done using the multiring PMUTs Characterization bonded to a PCB and using a network The electrical characterization analyzer in the air was(Agilent Technologies, done using the multiringSanta Clara, PMUTs CA, bonded USA). Figure 6a left inset shows a schematic of the experimental set-up. Each to a PCB and using a network analyzer (Agilent Technologies, Santa Clara, CA, USA). ring was powered Figure 6awith 10 dBm left inset showscontinuous wave a schematic to obtain of the the S-parameters. experimental set-up. EachThe ringfrequency re- was powered sponse, S 11, for each ring, gives a center value of 17.5 MHz, which corresponds to the res- with 10 dBm continuous wave to obtain the S-parameters. The frequency response, S11 , for onance frequency each ring, of an individual gives a center PMUT value of 17.5 MHz,(single whichclamped square corresponds PMUT, to the as reported resonance in frequency [26]). of an individual PMUT (single clamped square PMUT, as reported in [26]). (a) (b) Figure 6. Scattering parameter measurements (a) in air (S11 , red and blue curves with left axis; S12 , green and rose curves with right axis) using rings #4 and #5 and (b) in Fluorinert, S12 using rings #4 and #5. Sensors 2021, 21, 4786 9 of 18 In Figure 6a, the red curve shows the frequency response corresponding to ring #4, and the blue curve shows the ring #5 response. The multiple peaks are a consequence of the multiple individual PMUTs forming the ring. In order to analyze the electrical crosstalk effect in the proposed device, the S21 magnitude between pairs of rings was obtained [27]. In Figure 6a, the dotted graph (green curve) corresponds to the measurements between ring #5 and ring #4. The crosstalk level was obtained considering the S21 at 19 MHz (out of the resonance peak), giving for this case −59.4 dB which represents 2.1 mVpp (for a 10 dBm input signal). Even if the crosstalk level during the resonance is considered, there would be no significant actuation voltage in the non-actuated rings, decreasing the risk of acoustic interferences. Table 3 summarizes the obtained crosstalk between rings. Note that the lowest level, −77 dB, is between ring #1 and ring #4 and not between the most widely spaced rings (ring #1 and ring #5). This can be attributed to the specific layout of the electrical pads (consecutive pads for rings #1 and #5). The low crosstalk levels (below −56.8 dB) between multielement rings will allow driving each ring independently, ensuring a well-controlled and efficient axial beam-focusing. Table 3. Crosstalk between rings, according to S21 experimental magnitude at 19 MHz in air. dB 1 2 3 4 5 1 × −56.8 −76 −77 −70 2 × −69.5 −74 −72.5 3 × −62 −69.5 4 × −59.4 5 × The variation of resonance frequencies for each of the individual resonators due to mismatching during the fabrication can be a drawback. This problem can be alleviated when the system is under liquid operation due to the acoustic radiation mass loading effect from the liquid which widens the resonance frequency curve. For a square PMUT transducer under liquid operation in one side, this mass load damping can be quantified according to Equation (5) (where ρliquid is the liquid density, a is the transducer side, and µ is the mass per unit area) [28], which gives β = 2.46. ρliquid × a β = 0.342 (5) µ Then, the expected resonance frequency in liquid will be approximately half the air √ resonance frequency, i.e., 9.4 MHz (fair/ (1 + β)). Figure 6b shows the electrical frequency response of the crosstalk between ring #5 and ring #4 in FC-70. The resonance peaks appear between 8 and 10 MHz, being in correspondence with the expected value (9.4 MHz). On the other hand, in liquid, the high acoustic radiation damping is translated into a low quality factor and consequently to higher fractional bandwidth, hence lowering the S21 magnitude at resonance and smoothing the single-element frequency peaks of the same ring. 3.2. Output Pressure Measurements The multielement ring array was immersed in Fluorinert (c = 685 m/s, ρ = 1940 kg/m3 ) and each ring was driven by a signal generator (Keysight, Santa Rosa, CA, USA) with four sine cycles with 24 Vpp. The acoustic pressure was measured with a commercial hy- drophone from ONDA (Santa Clara, CA, USA) and displayed in an oscilloscope (Santa Rosa, CA, USA); Figure 7 shows the set-up. The experimental resonance frequency (tuned to maximize hydrophone signal) was 8.7 MHz in accordance with the electrical measurements. sine cycles with 24 Vpp. The acoustic pressure was measured with a commercial hydro- phone from ONDA (Santa Clara, CA, USA) and displayed in an oscilloscope (Santa Rosa, Sensors 2021, 21, 4786 10 of to CA, USA); Figure 7 shows the set-up. The experimental resonance frequency (tuned 18 maximize hydrophone signal) was 8.7 MHz in accordance with the electrical measure- ments. (a) (b) Figure 7. Set-up for for the the acoustic acousticcharacterization characterizationasasactuator actuatorininliquid liquid environment environment (a)(a) schematic schematic set-up set-up andand (b) (b) photo photo of of the the experimental experimental set-up. set-up. In order to avoid In avoid artifacts artifacts on onthe thepressure pressurefieldfieldcharacterization characterizationdue duetotospatial spatialaverag- aver- ing effects aging effects(consequence (consequence of the influence of the of the influence of hydrophone’s the hydrophone’s diameter), therethere diameter), is a limitation is a lim- on the on itation minimum the minimumaxial distance between axial distance the hydrophone between the hydrophone and theandultrasound transducer. the ultrasound trans- Equation ducer. (6) should Equation be usedbe[29] (6) should usedto estimate the maximum [29] to estimate effective the maximum hydrophone effective radius hydrophone (ah ), taking radius into account (ah), taking the transducer into account radius radius the transducer (a1 ), the(awavelength in the acoustic 1), the wavelength media in the acoustic (λ = c/f ), and the distance between the hydrophone and the transducer media (λ = c/f), and the distance between the hydrophone and the transducer (l). In our (l). In our experi- mental set-up,set-up, experimental ah =100 =100(HNC-0200 ah µm µm (HNC-0200 radius), c = 685 radius), c =m/s, 685 m/s, f = 8.69 and and MHz, f = 8.69 giving MHz, the giving the l/2a1l/2a ratioratio = 5.06, which 1 = 5.06, means which different means minimum different minimum distances depending distances dependingon the onring the size; ring TableTable size; 4 summarizes 4 summarizesthesethese values for allfor values cases (Lmin(Lmin all cases represents l in Equation represents (6)). (6)). l in Equation λλ ( ) q  aha= = 2 l l ++aa 2 2 2 (6) h 8a1 11 (6) 8a1 Therefore, in the experiments, the hydrophone was placed at the distances detailed Therefore, in the experiments, the hydrophone was placed at the distances detailed in Table 4 and raised every 50 µm to obtain the axial pressure at different heights without in Table 4 and raised every 50 µm to obtain the axial pressure at different heights without distortion. These distances are in the far-field, which allows defining the acoustic pressure distortion. These distances are in the far-field, which allows defining the acoustic pressure by Equation (7) [18] (where R0 is the Rayleigh distance, P0 is the surface pressure, and z is by theEquation (7) [18] (where R0 is the Rayleigh distance, P0 is the surface pressure, and z is axial distance). the axial distance). P0 R0 P(z) = (7) Pz0 R0 P(z) = (7) z Table 4. Individual ring characterization as actuator in FC-70 using HNC-0200 from ONDA (at 8.69 MHz). Lmin is computed from Equation (6); NP is experimentally measured. Table 4. Individual ring characterization as actuator in FC-70 using HNC-0200 from ONDA (at 8.69 MHz). Lmin is computed from Equation (6); NP is experimentally measured. Pressure (kPa) Ring Lmin (mm) NP (kPa×mm) at 1.2 mm Pressure (kPa) Ring 1 Lmin (mm) 0.531 NP (kPa*mm) 6.33 5.28 2 1.19 10.04 at 1.2 mm8.37 13 0.5311.84 6.33 14.92 5.2812.43 24 1.19 2.2 10.04 15.41 8.3712.84 35 1.843.16 14.92 14.2 12.4311.79 The measured points were fitted according to Equation (7), obtaining from the slope the normalized pressure with the distance NP = P0 R0 (Pa×mm) for each ring (see the results in Table 4). From these measurements, the maximum attainable pressure at the natural focus (~1.2 mm according to Field II simulations at 8.69 MHz) can be computed, Sensors 2021, 21, 4786 11 of 18 evaluating the NP for each ring at 1.2 mm and adding them because, at this point, all acoustic waves are in phase, allowing constructive interference. Finally, the computed total pressure is 50.71 kPapp . This output pressure could be increased until 4.8 times, for example, focusing at 200 µm (see the improvement factor in Figure 5a). The normalized acoustic pressure (ST) at 1.5 mm from a 1 mm2 PMUT array area when it is driven with 1 V was used to compare the multielement ring array with other ultrasound transducers. Taking the previously computed pressure when all rings are excited, 50.71 kPapp , and normalizing with the distance (1.2 mm) and the applied voltage (24 Vpp), we obtain a surface pressure of 2.54 kPapp ×mm×V−1 . According to this and considering the entire area of the multielement ring array (π × (D5 2 − d1 2 ) = 0.35 mm2 ), the normalized pressure at 1.5 mm is 4.84 kPa/V/mm2 . Table 5 compares the multielement ring array performance as an actuator with the state-of-the-art approaches, demonstrating a promising performance with a minute area. The normalized output pressure is 55% higher than that reported for arrays of annular AlN flexural rings (2.2 kPa/V/mm2 using five- channel ring [6]) and even 92% higher than that for a system for intravascular imaging [2]. Table 5. Comparison of ring arrays with different technological approaches. [30] [2] [6] Parameters This Work 2011 2018 2019 Transducer CMUT PZT matrix AlN PMUT AlN PMUT technology Configuration Multielement ring Multielement ring Continuous ring Multielement ring Medium Vegetable oil Water Mineral oil FC-70 Frequency (MHz) 1.2 14 6 8.69 Area (mm2 ) 12.76 1 6.28 1 7.07 1 0.35 Pressure 13.2 0.4 2.8 2.11 (kPa/V@mm) @1.5 2 @6 @5.4 @1.2 NP (kPa×mm/V) 19.8 2.4 15.2 2.54 ST (kPa/V/mm2 ) 1.11 3 0.38 3 2.18 3 4.84 1Computed considering the transducers’ dimensions. 2 Taking the peak-to-peak pressure (2 × 66 kPa) and the applied voltage (10 V). 3 Computing as the ratio between NP evaluated at 1.5 mm and the area. In order to complete a deeper characterization of the performance of the system as an actuator and compare the presented multielement array system with a continuous AlN flexural PMUT array [6], the pressure normalized with the area and energy density, (p), defined in Equation (8) was computed (where V is the applied voltage (24 V), Aap is the area (0.35 mm2 ), e31,f is the piezoelectric coefficient (e31,f = e31 − υ × e33 = −1.065 C/m2 ), λ is the wavelength (λ = c/f = 79.86 µm), and p is the total pressure (50.71 kPa at 1.2 mm)). λ p = p× (8) V × A ap × e31, f Applying this expression at 1.2 mm (the natural focus) when the multielement five-ring array is used, a normalized pressure per area per energy density of 453 kPa×mm−2 /J×cm−3 is obtained, which is 2.5× better than that of the five-channel continuous ring array without delays presented in [6] (184 kPa×mm−2 /J×cm−3 ). Compared with the result when the same system is focused at 1.9 mm (588 kPa×mm−2 /J×cm−3 , [6]), our multielement ring array still exhibits a competitive value and will allow higher pressure output at shorter focus depths if some phase-beam focusing is used. 3.3. Pulse-Echo Measurements Thanks to its reduced area, the multielement ring array can be a candidate for catheter- based ultrasound imaging; consequently, it was also characterized as a pulse-echo acoustic system. In this case, the central ring was used as a receiver, and the rings #5, #4, and #3 were used as transmitters. The same signal generator was used to drive the transmission with four sine cycles at 8.69 MHz with 24 Vpp. The receiving ring was externally connected 3.3. Pulse-Echo Measurements Thanks to its reduced area, the multielement ring array can be a candidate for cathe- ter-based ultrasound imaging; consequently, it was also characterized as a pulse-echo Sensors 2021, 21, 4786 12 of 18 acoustic system. In this case, the central ring was used as a receiver, and the rings #5, #4, and #3 were used as transmitters. The same signal generator was used to drive the trans- mission with four sine cycles at 8.69 MHz with 24 Vpp. The receiving ring was externally connected to an integrated to an integrated CMOSamplifier CMOS voltage voltage amplifier with with a gain of a25gain dB of 25 The [31]. dB [31]. liquidThe liquid thickness thickness over the PMUT chip was tuned to obtain different times of flight over the PMUT chip was tuned to obtain different times of flight (ToFs) and consequently(ToFs) and consequently different different acoustic acoustic paths paths (AP = ToF × c)(AP = ToF giving * c) giving a round a round trip from 1 to trip from 5 mm; 1 to 85 shows Figure mm; Figure 8 shows the set-up. the set-up. (a) (b) Figure 8.8. Figure (a)(a) Schematic set-up Schematic forfor set-up pulse-echo measurements; pulse-echo measurements;(b) (b)photo photoofofthe theexperimental experimentalset-up. set-up. Figure9 9shows Figure showsthethemeasured measuredecho echovoltage voltageandanditsitsdependence dependenceononthe theacoustic acousticpath path using ring #3 (blue points) (same measurements were also acquired using ring #3 (blue points) (same measurements were also acquired for ring #4 and ringfor ring #4 and ring #5). Taking the fitting results, the receiving sensitivity (SR) can be computed #5). Taking the fitting results, the receiving sensitivity (SR) can be computed as Rx/Pt as Rx/Pt (where Pt is defined (where by the by Pt is defined normalized pressure, the normalized NP/z, in pressure, NP/z,Tablein4), giving Table 4), an average giving value of an average 441.6 nV/Pa ((Rx value of 441.6 nV/Pa /Pt ((Rxcentral central + Rx 3 /Pt3+ central /Pt + Rx Rxcentral/Pt4 4+ Rxcentral /Pt )/3). /Pt5)/3). central This receiving sensitivity 5 This receiving sensitivity is is affected affected byparasitic by the the parasitic capacitances capacitances betweenbetween the PMUT the PMUT and voltage and voltage amplifier amplifier (PCB, (PCB, con- connectors, cables, etc.) according to Equation nectors, cables, etc.) according to Equation (9) [32]: (9) [32]: CcentralRing SR = SR EOC × G × (9) CcentralRing + CinLN A + C parasitic where SREOC is the “end-of-cable open-circuit sensitivity”; G is the amplifier gain (25 dB); CcentralRing is the receiving element capacitance (254.9 fF extracted from COMSOL); CinLNA is the voltage amplifier equivalent capacitance (609 fF) [16]; and Cparasitic is the parasitic capacitance associated with PCBs, connectors, cables, etc. (at 6.5 pF). Hence, the intrinsic sensitivity (SREOC ) for the smaller inner ring is 717.8 nV/Pa. This value is very competitive in comparison with some commercial hydrophones with sizes comparable to our ring #1 (diameter of ~149 µm) but with smaller nominal SREOC pressure sensitivities: HNC- 0200 from ONDA (200 µm diameter, SREOC = 28 nV/Pa [32]) or NH0200 from Precision Acoustics (200 µm, 55 nV/Pa with amplifier [33]). The time-domain response shown as the red curve in Figure 10 corresponds to pulse- echo measurement when ring #4 is used to transmit and the central ring is used to receive. The time-of-flight of the received echo is 2.34 µs, which gives an FC-70 thickness of 800 µm (FC70thickness = ToF × c/2). Taking the ringdown, the fast Fourier transform was computed (see Figure 10, blue curve), giving a resonance frequency (f0 ) of 8.8 MHz with fractional bandwidth at −6 dB close to 54%. the voltage amplifier equivalent capacitance (609 fF) [16]; and Cparasitic is the parasitic ca- pacitance associated with PCBs, connectors, cables, etc. (at 6.5 pF). Hence, the intrinsic sensitivity (SREOC) for the smaller inner ring is 717.8 nV/Pa. This value is very competitive in comparison with some commercial hydrophones with sizes comparable to our ring #1 Sensors 2021, 21, 4786 (diameter of ~149 µm) but with smaller nominal SREOC pressure sensitivities: HNC-0200 13 of 18 from ONDA (200 µm diameter, SREOC = 28 nV/Pa [32]) or NH0200 from Precision Acoustics (200 µm, 55 nV/Pa with amplifier [33]). Figure 9. 9. Pulse-echo experiment with ring #3 as transmitter transmitter and ring #1#1 as as receiver. receiver. The The received received signal is plotted as a function of the acoustic path, AP, using FC-70-air interface as reflecting surface. signal is plotted as a function of the acoustic path, AP, using FC-70- air interface as reflecting surface. Sensors 2021, 21, x FOR PEER REVIEW Error bar bar == ±200 14 of 18 Error ±200µV. µV. Inset: Inset: Time-domain Time-domain signal signal at at 2.8 2.8 mm mm round round trip. trip. The time-domain response shown as the red curve in Figure 10 corresponds to pulse- echo measurement when ring #4 is used to transmit and the central ring is used to receive. The time-of-flight of the received echo is 2.34 µs, which gives an FC-70 thickness of 800 µm (FC70thickness = ToF*c/2). Taking the ringdown, the fast Fourier transform was computed (see Figure 10, blue curve), giving a resonance frequency (f0) of 8.8 MHz with fractional bandwidth at −6 dB close to 54%. 10. Acoustic pulse-echo measurement using ring #4 to transmit Figure 10. transmit and central central ring ring to to receive receive (FC-70 thickness (FC-70 thickness at 800μm). at800 µm).Red Redcurve (top-right curve redred (top-right axis): time-domain axis): response. time-domain Blue Blue response. curvecurve (left- bottom blue axis): FFT from the ringdown. (left-bottom blue axis): FFT from the ringdown. The acoustic beamwidth was determined through aa pulse-echo pulse-echo experiment experiment where where ring #3 and ring #4 were were used as as transmitters transmitters and the the central central ring ring was was used used as as the the receiver. receiver. µm diameter A 150 µm diameter conductive conductive wire was was used used as as aa reflecting reflecting surface surface and and was was placed placed at at 790 µm over the array’s surface. The wire covered all PMUTs PMUTs in in one one lateral lateral direction, direction, and and in the other direction, it was mechanically displaced 500 µm to each side from the the center center of the array (see Figure 11 inset). Figure 11 provides experimental experimental andand simulation simulation results results (using Field II) for for when when only only ring ring #3#3 or or ring ring #3 #3 ++ ring ring #4 #4 were were excited. excited. According According to to the the simulations, the beamwidth at − −66 dB dB is is around around 160160 µmµm when ring #3 + ring #4 are used used (blue lines in Figure 11) and is 180 µmµm when only ring #3 is used (red lines in Figure 11). Moreover, the maximum amplitude increases by about 6 dB when both rings are used. The experimental points demonstrate a very good agreement with the simulated ones. A 150 µm diameter conductive wire was used as a reflecting surface and was placed a 790 µm over the array’s surface. The wire covered all PMUTs in one lateral direction, an in the other direction, it was mechanically displaced 500 µm to each side from the cente of the array (see Figure 11 inset). Figure 11 provides experimental and simulation result Sensors 2021, 21, 4786 14 of 18 (using Field II) for when only ring #3 or ring #3 + ring #4 were excited. According to th simulations, the beamwidth at −6 dB is around 160 µm when ring #3 + ring #4 are use (blue lines in Figure 11) and is 180 µm when only ring #3 is used (red lines in Figure 11 Moreover, Moreover, thethe maximum maximum amplitude amplitude increases increases by aboutby6 dB about when6 both dB when both rings are rings used. Theare used experimental points demonstrate a very good agreement with the simulated ones. The experimental points demonstrate a very good agreement with the simulated ones. Figure 11. Simulated and experimental behavior along the lateral direction at 790 µm axial distance. Red curves correspond to ring #3 as transmitter and blue curves correspond to ring #3 + ring #4 (dots: experimental points; lines: Field II simulations). Inset: Experimental set-up. 3.4. Focusing Capabilities The focusing capabilities were demonstrated through a pulse-echo experiment using ring #3 and ring #5 as transmitters and the central ring as a receiver (see set-up in Figure 8). The acoustic path was tuned in the same way, changing the FC-70 thickness and measuring the time-of-flight corresponding to ring #3 (nearest to receiver element). Figure 12 (inset) shows the signal received by the central ring when ring #3 and ring #5 are used separately. As can be seen, the received acoustic waves from ring #3 and ring #5 are not in phase; in consequence, when both rings are driven together, the total acoustic pressure is less than the sum of both echoes. The behavior without any delays is shown in Figure 12 (green points), giving a maximum level of 6.59 mVpp close to 1.5 mm round trip (natural focus). Considering this value and the computed receiving sensitivity (441.6 nV/Pa), the pressure on the array’s surface is 14.9 kPapp . For electronic focusing or phased-array rings, 34 ns was applied to ring #3 during transmission to allow both acoustic waves to arrive in phase, achieving a maximum acoustic pressure at 0.6 mm (see Figure 12, orange points). Computing the focusing improvement factor as the ratio between maximum received echoes (8.4 mVpp/6.59 mVpp), a 1.3× improvement factor is obtained, which is translated into an acoustic pressure of 19.4 kPapp (1.3 × 14.9 kPapp ). For electronic focusing or phased-array rings, 34 ns was applied to ring #3 during transmission to allow both acoustic waves to arrive in phase, achieving a maximum acous tic pressure at 0.6 mm (see Figure 12, orange points). Computing the focusing improve ment factor as the ratio between maximum received echoes (8.4 mVpp/6.59mVpp), a 1.3x Sensors 2021, 21, 4786 improvement factor is obtained, which is translated into an acoustic pressure15of of 18 19.4 kPapp (1.3 × 14.9 kPapp). Figure 12. Voltage measured by the central ring when ring #3 and ring #5 are driven at the same time Figure 12. Voltage (green points) measuredfocused or electronically by the (orange central points). ring when ring Inset: #3 and ring Time-domain #5 are of response driven at the same the echo time signal(green points) from ring #3 andorring electronically focused #5 to determine (orange the needed timepoints). Inset: delay (34 ns). Time-domain response of the echo signal from ring #3 and ring #5 to determine the needed time delay (34 ns). 3.5. 1D Line-Scan Imaging 3.5. 1D TheLine-Scan Imaging was tested using a Cu grating phantom with three holes with imaging capability different widths (600, 900, and 1040 µm) and gaps (1.2 and 1 mm) between them (see The imaging capability was tested using a Cu grating phantom with three holes with Figure 13 inset). The sample was immersed in FC-70 and placed at 790 µm on top of the different widths (600, 900, and 1040 µm) and gaps (1.2 and 1 mm) between them (see Fig- multielement ring array. A micrometric system was used to displace it along the x-direction ure 13 inset). The (perpendicular to thesample was steps holes) with immersed in FC-70 of 50 µm, andy-direction while the placed atwas790fixed µm on top of the at the multielement sample’s center. Two rings (#3 and #4) were driven with four cycles at 8.69 MHz with x-direc ring array. A micrometric system was used to displace it along the tion (perpendicular 24 Vpp, and the centraltoring the(connected holes) with steps to the CMOSof 50 µm, amplifier voltage while theasy-direction was fixed a before) was used the sample’s to detect center. echoes the reflected Two rings (#3set-up (see the and #4) were driven in Figure 8). with four cycles at 8.69 MHz with Figure 13 (red points) show the experimental peak-to-peak amplitudes received by the central ring, giving a maximum value of around 7.5 mVpp and clearly reproducing the AA’ profile with the three holes. On the other hand, a scatter phantom close to the grating was modeled in Field II with the purpose of obtaining the scanning pattern under the same assumptions. The normalized signal received by the central ring when rings #3 and #4 are excited is shown in Figure 13 (blue points), demonstrating a good agreement with the experimental one and validating the capability to perform acoustic imaging with the multielement ring PMUT presented. On the other hand, a scatter phantom close to the grating was modeled in Field II with the purpose of obtaining the scanning pattern under the same assumptions. The nor- malized signal received by the central ring when rings #3 and #4 are excited is shown in Figure 13 (blue points), demonstrating a good agreement with the experimental one and Sensors 2021, 21, 4786 validating the capability to perform acoustic imaging with the multielement ring PMUT 16 of 18 presented. Figure 13. Pulse-echo ultrasonic 1D image of a grating phantom. Red curve: experimental points; Figure 13. Pulse-echo ultrasonic 1D image of a grating phantom. Red curve: experimental points; blue curve: Field II simulations. Inset: Grating phantom with A-A’ scanned profile. blue curve: Field II simulations. Inset: Grating phantom with A-A’ scanned profile. 4. Conclusions 4. Conclusions This paper presents a multielement ring ultrasound transducer array based on AlN This paper presents PMUTs fabricated with a aMEMS-on-CMOS multielement ring ultrasound process. transducer The presented array based multielement on array ring AlN PMUTs fabricated with a MEMS-on-CMOS process. The presented eliminates the dependence of acoustic wave frequency on the diameters of the annular multielement ring ar- ray eliminates the dependence of acoustic wave frequency on array and achieves accurate control of the focus depth (from 67 µm to 1.4 mm), whichthe diameters of the annular array is 4.9 and timesachieves greateraccurate than thatcontrol of the of the focuscontinuous equivalent depth (fromring 67 µm to 1.4 array. Themm), which is low crosstalk 4.9 times different between greater than ringsthat of the (levels equivalent under −56.8 dB) continuous allows itring to bearray. The used in low crosstalk modern be- ultrasound tween different applications whereringsthe(levels maximum underof−56.8 dB) allowsbeam the ultrasound it to must be used in modernefficiently be controlled ultrasound in applications where The the axial direction. the maximum PMUT-based of the ringultrasound array, withbeama verymust be controlled reduced efficiently area, generates high in the axial pressure direction. levels (4.84 kPa/V/mm 2 at 1.5 mm) The PMUT-based ringatarray, 8.7 MHz within a liquid very reduced area, being environment, generates very high pressure levels (4.84 kPa/V/mm 2 at 1.5 mm) at 8.7 MHz in a liquid environment, be- competitive with other annular arrays using bulk piezoelectric, CMUT, or PMUT fabrication ing very competitive approaches. with other The pulse-echo annular arrays experiments with ausing voltagebulk piezoelectric, amplifier (gain ofCMUT, or PMUT 25 dB) externally fabrication connected to approaches. the centralThe ringpulse-echo gave a receivingexperiments withof sensitivity a 441.6 voltage amplifier nV/Pa, which(gain of 25 could be dB) externally increased aroundconnected 700 nV/Pa to the central when ring gave the PMUT a receiving sensitivity is monolithically integrated ofon441.6 the nV/Pa, CMOS circuitry. which Thebe1D could imagingaround increased test through 700 nV/Pamechanical when the scanning PMUT demonstrates is monolithicallythe integrated possibility to obtain high-performance ultrasound imaging systems. on the CMOS circuitry. The 1D imaging test through mechanical scanning demonstrates With this performance and considering its small size (below 1 mm 2 ), the presented multielement ring array fabricated the possibility to obtain high-performance ultrasound imaging systems. With this perfor- with a and mance PMUT-on-CMOS considering itstechnology becomes small size (below 1 mman 2interesting ), the presentedultrasound transducer multielement for ring ar- applications ray fabricatedinwith whicha size, cost, reliability,technology PMUT-on-CMOS and performance becomes arean a must, such asultrasound interesting wearables and catheter-based systems. Greater focal depth and output pressure can be achieved at the same frequency by increasing the number of rings within the same technology. Author Contributions: Conceptualization, E.L., A.U., and N.B.; device design, simulations, and acoustic experiments, E.L.; validation and analysis, E.L., I.Z., and N.B.; writing—original draft prepa- ration, E.L.; writing—review and editing, I.Z., A.U., and N.B.; supervision, A.U. and N.B.; project administration, N.B. All authors have read and agreed to the published version of the manuscript. Funding: This research was partially funded by the Spanish AEI with project PID2019-108270RB-I00. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable. Sensors 2021, 21, 4786 17 of 18 Acknowledgments: Special acknowledges to SilTerra’s MEMS and SENSORS technology develop- ment team for support in fabricating the PMUT-on-CMOS. Conflicts of Interest: The authors declare no conflict of interest. References 1. Dangi, A.; Agrawal, S.; Tiwari, S.; Jadhav, S.; Cheng, C.; Datta, G.R.; Troiler-McKinstry, S.; Pratap, R.; Kothapalli, S.-R. Ring PMUT array based miniaturized photoacoustic endoscopy device. In Proceedings of the Photons Plus Ultrasound: Imaging and Sensing 2019, San Francisco, CA, USA, 3–6 February 2019; p. 1087811. 2. Janjic, J.; Tan, M.; Daeichin, V.; Noothout, E.; Chen, C.; Chen, Z.; Chang, Z.Y.; Beurskens, R.H.S.H.; Van Soest, G.; Van Der Steen, A.F.W.; et al. A 2-D Ultrasound Transducer With Front-End ASIC and Low Cable Count for 3-D Forward-Looking Intravascular Imaging: Performance and Characterization. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2018, 65, 1832–1844. [CrossRef] 3. Moini, A.; Nikoozadeh, A.; Choe, J.W.; Chang, C.; Stephens, D.N.; Sahn, D.J.; Khuri-Yakub, P.T. Fully integrated 2D CMUT ring arrays for endoscopic ultrasound. In Proceedings of the IEEE International Ultrasonics Symposium, IUS, Tours, France, 18–21 September 2016; pp. 9–12. 4. Zahorian, J.; Hochman, M.; Xu, T.; Satir, S.; Gurun, G.; Karaman, M.; Degertekin, F.L. Monolithic CMUT-on-CMOS integration for intravascular ultrasound applications. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2011, 58, 2659–2667. [CrossRef] [PubMed] 5. Bawiec, C.R.; N’Djin, W.A.; Bouchoux, G.; Sénégond, N.; Guillen, N.; Chapelon, J.Y. Preliminary Investigation of a 64-element Capacitive Micromachined Ultrasound Transducer (CMUT) Annular Array Designed for High Intensity Focused Ultrasound (HIFU). IRBM 2018, 39, 295–306. [CrossRef] 6. Eovino, B.E.; Liang, Y.; Lin, L. Concentric PMUT Arrays for Focused Ultrasound and High Intensity Applications. In Proceedings of the 2019 IEEE 32nd International Conference on Micro Electro Mechanical Systems (MEMS), Seoul, Korea, 27–31 January 2019; pp. 771–774. 7. Li, G.; Qiu, W.; Zhang, Z.; Jiang, Q.; Su, M.; Cai, R.; Li, Y.; Cai, F.; Deng, Z.; Xu, D.; et al. Noninvasive Ultrasonic Neuromodulation in Freely Moving Mice. IEEE Trans. Biomed. Eng. 2019, 66, 217–224. [CrossRef] [PubMed] 8. Kim, H.; Kim, S.; Sim, N.S.; Pasquinelli, C.; Thielscher, A.; Lee, J.H.; Lee, H.J. Miniature ultrasound ring array transducers for transcranial ultrasound neuromodulation of freely-moving small animals. Brain Stimul. 2019, 12, 251–255. [CrossRef] [PubMed] 9. Hosseini, S.; Laursen, K.; Rashidi, A.; Mondal, T.; Corbett, B.; Moradi, F. S-MRUT: Sectored-Multiring Ultrasonic Transducer for Selective Powering of Brain Implants. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2021, 68, 191–200. [CrossRef] [PubMed] 10. Weber, M.J.; Yoshihara, Y.; Sawaby, A.; Charthad, J.; Chang, T.C.; Arbabian, A. A Miniaturized Single-Transducer Implantable Pressure Sensor with Time-Multiplexed Ultrasonic Data and Power Links. IEEE J. Solid-State Circuits 2018, 53, 1089–1101. [CrossRef] 11. Shi, C.; Andino-Pavlovsky, V.; Lee, S.A.; Costa, T.; Elloian, J.; Konofagou, E.E.; Shepard, K.L. Application of a sub-0.1-mm3 implantable mote for in vivo real-time wireless temperature sensing. Sci. Adv. 2021, 7, 1–10. [CrossRef] 12. Wang, J.; Zheng, Z.; Chan, J.; Yeow, J.T.W. Capacitive micromachined ultrasound transducers for intravascular ultrasound imaging. Microsyst. Nanoeng. 2020, 6, 73. [CrossRef] 13. Wang, D.; Filoux, E.; Levassort, F.; Lethiecq, M.; Rocks, S.A.; Dorey, R.A. Fabrication and characterization of annular-array, high-frequency, ultrasonic transducers based on PZT thick film. Sensors Actuators A Phys. 2014, 216, 207–213. [CrossRef] 14. Brenner, K.; Ergun, A.S.; Firouzi, K.; Rasmussen, M.F.; Stedman, Q.; Khuri-Yakub, B. Advances in capacitive micromachined ultrasonic transducers. Micromachines 2019, 10, 152. [CrossRef] 15. Jiang, X.; Lu, Y.; Tang, H.-Y.; Tsai, J.M.; Ng, E.J.; Daneman, M.J.; Boser, B.E.; Horsley, D.A. Monolithic ultrasound fingerprint sensor. Microsyst. Nanoeng. 2017, 3, 17059. [CrossRef] 16. Zamora, I.; Ledesma, E.; Uranga, A.; Barniol, N. Monolithic Single PMUT-on-CMOS Ultrasound System with +17 dB SNR for Imaging Applications. IEEE Access 2020, 8, 142785–142794. [CrossRef] 17. Lu, Y.; Heidari, A.; Horsley, D.A. A High Fill-Factor Annular Array of High Frequency Piezoelectric Micromachined Ultrasonic Transducers. J. Microelectromechanical Syst. 2015, 24, 904–913. [CrossRef] 18. Blackstock, D.T. Fundamentals of Physical Acoustics; John Wiley & Sons: New York, NY, USA, 2000; ISBN 0471319791. 19. Blevins, R.D. Formulas for Natural Frequency and Mode Shape; Van Nostrand Reinhold Co., Ed.; Litton Educational Publishing, Inc.: New York, NY, USA, 1979. 20. Horsley, D.; Lu, Y.; Rozen, O. Flexural Piezoelectric Resonators. In Piezoelectric MEMS Resonators; Bhugra, H., Piazza, G., Eds.; Springer International: Cham, Switzerland, 2017; pp. 153–173, ISBN 9783319286884. 21. Ledesma, E.; Zamora, I.; Uranga, A.; Barniol, N. Tent-plate AlN PMUT with a piston-like shape under liquid operation. IEEE Sens. J. 2020, 20, 11128–11137. [CrossRef] 22. Amabili, M.; Frosali, G.; Kwak, M.K. Free vibrations of annular plates coupled with fluids. J. Sound Vib. 1996, 191, 825–846. [CrossRef] 23. Jensen, J.A.; Svendsen, N.B. Calculation of pressure fields from arbitrarily shaped, apodized, and excited ultrasound transducers. Ultrason. Ferroelectr. Freq. Control. IEEE Trans. 1992, 39, 262–267. [CrossRef] [PubMed] 24. Jensen, J.A. Field: A program for simulating ultrasound systems. Med. Biol. Eng. Comput. 1996, 34, 351–352. 25. Przybyla, R.J.; Tang, H.; Member, S.; Guedes, A.; Shelton, S.E.; Horsley, D.A.; Boser, B.E. 3D Ultrasonic Rangefinder on a Chip. IEEE J. Solid-State Circuits 2015, 50, 320–334. [CrossRef] Sensors 2021, 21, 4786 18 of 18 26. Ledesma, E.; Zamora, I.; Tzanov, V.; Torres, F.; Uranga, A.; Barniol, N.; Marigó, E.; Soundara-Pandian, M. Liquid operable AlN PMUT with high output pressure capabilities. In Proceedings of the 2019 IEEE International Ultrasonics Symposium (IUS), Glasgow, UK, 6–9 October 2019; pp. 251–254. 27. He, L.M.; Xu, W.J.; Liu, W.J.; Wang, X.B.; Zhou, J.; Ren, J.Y. Performance and Crosstalk Evaluation of 2-D Array Piezoelectric Micromachined Ultrasonic Transducer with 3-D Finite Element Simulation. In Proceedings of the IEEE International Ultrasonics Symposium, IUS, Glasgow, UK, 6–9 October 2019; pp. 792–795. 28. Bernstein, J.J.; Finberg, S.L.; Houston, K.; Niles, L.C.; Daniel Chen, H.; Eric Cross, L.; Li, K.K.; Udayakumar, K. Micromachined high frequency ferroelectric sonar transducers. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 1997, 44, 960–969. [CrossRef] 29. Smith, R.A. Are hydrophones of diameter 0.5 mm small enough to characterise diagnostic ultrasound equipment? Phys. Med. Biol. 1989, 34, 1593–1607. [CrossRef] [PubMed] 30. Cristman, P.; Oralkan, O.; Mandella, M.; Solgaard, O.; Contag, C.; Khuri-Yakub, B.T. Interdigitated annular CMUT arrays for ultrasound assisted delivery of fluorescent contrast agents. In Proceedings of the IEEE International Ultrasonics Symposium, IUS, Orlando, FL, USA, 18–21 October 2011; pp. 96–99. 31. Zamora, I.; Ledesma, E.; Uranga, A.; Barniol, N. Miniaturized 0.13-µm CMOS Front-End Analog for AlN PMUT Arrays. Sensors 2020, 20, 1205. [CrossRef] [PubMed] 32. Onda Corporation Hydrophone Handbook. Available online: http://ondacorp.com/Handbook/mobile/index.html#p=12 (accessed on 8 June 2021). [CrossRef] 33. Precision Acoustics Ltd. 0.2 mm Needle Hydrophone (NH0200). Available online: https://www.acoustics.co.uk/pal/wp- content/uploads/2016/05/nh0200-0.2mm-needle-hydrophone-tds.pdf (accessed on 8 June 2021).