sensors Article Metal Microelectromechanical Resonator Exhibiting Fast Human Activity Detection Francesc Torres * , Arantxa Uranga and Núria Barniol Electronic Engineering Department, Universitat Autònoma de Barcelona, Edifici Q, Campus UAB, 08193 Cerdanyola del Valles, Spain; arantxa.uranga@uab.cat (A.U.); nuria.barniol@uab.cat (N.B.) * Correspondence: francesc.torres@uab.cat; Tel.: +34-935868463 Abstract: This work presents a MEMS resonator used as an ultra-high resolution water vapor sensor (humidity sensing) to detect human activity through finger movement as a demonstrator example. This microelectromechanical resonator is designed as a clamped-clamped beam fabricated using the top metal layer of a commercial CMOS technology (0.35 µm CMOS-AMS) and monolithically integrated with conditioning and readout circuitry. Sensing is performed through the resonance frequency change due to the addition of water onto the clamped-clamped beam coming from the moisture created by the evaporation of water in the human body. The sensitivity and high-speed response to the addition of water onto the metal bridge, as well as the quick dewetting of the surface, make it suitable for low-power human activity sensing. Keywords: CMOS; humidity sensor; low power; MEMS; monolithic integration; resonators; non-contact detection 1. Introduction Human activity sensors have great importance for many applications, from movement and proximity sensing to breath monitoring; they are capable of offering valuable platforms Citation: Torres, F.; Uranga, A.; for human–machine non-contact interfaces and valuable insights into health conditions Barniol, N. Metal (i.e., respiratory rate and sleep depth to detect sleep apnea, or finger movement tapping Microelectromechanical Resonator rates for early Parkinson disease detection [1]). In parallel, high-performance humidity Exhibiting Fast Human Activity detection has taken an important development. Particularly, real-time humidity detection Detection. Sensors 2023, 23, 8945. allows us to design sensors for human activity such as breath monitoring or non-contact https://doi.org/10.3390/ finger movement [2–7]. s23218945 In both cases, the level of humidity is changed locally due to the water vapor contents Academic Editors: Edmond Cretu in any human exhalation during breathing, and the sharp gradient of humidity around the and Chang Ge fingertip [2] due to skin transpiration. Humidity sensors can provide a convenient and effective means of human activity Received: 28 September 2023 monitoring, with a non-intrusive, non-contact nature, and high sensitivity, making them a Revised: 24 October 2023 popular choice. Accepted: 26 October 2023 To detect humidity, there are many proposals in the literature, from capacitive [1] Published: 3 November 2023 or resistive sensors [8,9], surface acoustic waves (SAW) [10–12], or sensors based on me- chanical resonators [13–20], among others. Most of them require a specifically designed geometry, and some kind of functionalization using a sensitive layer or special materials Copyright: © 2023 by the authors. (2D materials [7], nanoforest structures [12], etc.) to increase the final achievable resolution. Licensee MDPI, Basel, Switzerland. As an example of resistive sensors, the authors of [9] present an interesting and easy- This article is an open access article to-fabricate flexible sensor based on cellulose with simple pencil-on-paper hand-drawn distributed under the terms and interdigitated electrodes. The paper is treated with a solution of NaCl, and a response time conditions of the Creative Commons of near 29 s is obtained when the sensor is cycled from 15% to 40% of relative humidity Attribution (CC BY) license (https:// (RH). This sensor is used for finger movement detection and respiratory monitoring, among creativecommons.org/licenses/by/ other uses. 4.0/). Sensors 2023, 23, 8945. https://doi.org/10.3390/s23218945 https://www.mdpi.com/journal/sensors Sensors 2023, 23, 8945 2 of 13 The authors of [10] present a very interesting sensor based on a SAW resonator with graphene oxide and MoS2 (as a sensing layer) films. Graphene oxide is used to prevent the short circuiting of the electrodes by a high conductive MoS2 film in high relative ambient humidity. The authors have obtained a high sensitivity of 12.41 kHz/%RH that corresponds to 0.005% of the change of frequency of resonance per %RH. One of the main uses of MEMS resonators (cantilevers, bridges, or plates) is as a gravimetric sensor that detects mass changes through the change of their resonant frequency. This frequency change can be sensed mainly using well-known optical, piezoelectrical, or capacitive transductions, among others. The capacitive transduction is widely used and is a perfect candidate to be monolithically integrated with a signal processing and readout circuit made with CMOS commercial technology. For example, in [17] the authors present a very interesting 4-anchored plate oscillator integrated monolithically with CMOS technology. The plate has dimensions of 41 µm × 10 µm with distributed holes to facilitate the releasing. The authors have obtained a sensitivity of 0.8% in frequency of resonance decreasing per %RH. In [18] the authors have proposed a piezoelectric cantilever (300 × 1000 µm2 ) with an MoS2 film deposited onto it, with a fast response time of 1 s, and 2.8 s of recuperation time and with a good sensitivity of 0.3% in RH. This cantilever is implemented with machine learning algorithms to perform the analysis of the results. The authors from [19] present a commercial double cantilever system, one that is 110 µm in length, and the other that is 50 µm in length, which is used as a reference. The authors use piezoresistive detection and obtain a mass sensitivity of near 300 fg/Hz. The sensor operates in a dynamical mode using a piezoelectric material to shake the cantilevers. The detection is performed by measuring both the frequency shift due to the water (mass) addition onto the cantilever, and the increase of the amplitude of vibration due to the water addition. In [20], a CMOS-MEMS humidity sensor is presented, based on a moving plate of 400 µm × 400 µm of surface, and using a CMOS technology TiO2 layer to absorb the water. The changes in the amplitude of vibration (detected using piezoresistive effect) are used for sensing. Also, the sensor is actuated using electrothermal transduction and is operated in dynamical mode. The authors have obtained a sensitivity of 0.1 mV/%RH. The metal sub-micrometric (clamped-clamped) cantilever resonator, which is inte- grated with CMOS technology, has been demonstrated to be a very sensitive mass sen- sor [14] and humidity sensor [15], acting as a mass sensor (water adsorption at the resonator surface). A comparison of this system with other systems based on optical and piezoelectric transduction was performed. Water absorption onto the clamped-clamped beam produces a change of the mass of the resonator, and consequently, a change of the resonance frequency occurs, following (1): f0 ∆f = − ∆m, (1) 2me f f where f 0 is the resonance frequency of the clamped-clamped beam, ∆m is the added mass, and meff is the effective mass of the clamped-clamped beam (0.4m0 where m0 is the mass of the bridge) [21]. Due to its low effective mass and high frequency of resonance, the sensor is capable of detecting humidity with a high sensitivity without the need of any functionalization layer. Due to this, we propose this MEMS resonator as a highly sensitive, non-contact and fast human activity detector. 2. Sensor Design and Fabrication The MEMS sensor utilized in this study is 18 µm long, 600 nm wide, and an 850 nm thick metal bridge resonator. It was fabricated using the top metal layer of a commercial 0.35 µm CMOS technology provided by Austria Microsystems (AMS) (Premstaetten, Austria) (3.3 V, 4-metal 2-poly), as shown in Figure 1. For the top metal we have used the 2. Sensor Design and Fabrication The MEMS sensor utilized in this study is 18 µm long, 600 nm wide, and an 850 nm Sensors 2023, 23, 8945 3 of 13 thick metal bridge resonator. It was fabricated using the top metal layer of a commercial 0.35 µm CMOS technology provided by Austria Microsystems (AMS) (Premstaetten, Aus- tria) (3.3 V, 4-metal 2-poly), as shown in Figure 1. For the top metal we have used the minimum width minimum possible width offeredpossible offered by this by this technology technology (0.6 µm). The (0.6design µm). The of design of the system does the system not incorporate does not incorporate a passivealayer, passive layer, enabling enabling a straightforward a straightforward post-CMOS post-CMOS process process with- without the out the need of need anyofmask. any mask. This process This process involves involves the wet the wet etching etching of the of the intermetal intermetal oxide (sacrificial oxide (sacrificial layer) to release the bridge, immersing the chip into hydrofluoric acid for a fewa few seconds, layer) to release the bridge, immersing the chip into hydrofluoric acid for and then seconds, and then rinsing rinsing it with it with deionized deionized water water and and isopropyl isopropyl alcohol alcohol to avoid to avoid possible stiction of possible the beam into electrodes or bottom layers. Since the resonator stiction of the beam into electrodes or bottom layers. Since the resonator is located on the is located on the top metal top metal layer,layer, thisthis release release process process is expedited, is expedited, allowing allowing forfor thethe maximum maximum preservation of the metal preservation by minimizing its etching rate compared to that of of the metal by minimizing its etching rate compared to that of the intermetal oxide.the intermetal oxide. TheThe resonator resonator is capacitively connected to a CMOS readout amplification circuitry through the the sensing is capacitively connected to a CMOS readout amplification circuitry through electrode sensing electrode (transimpedance (transimpedance amplification amplification withwitha gaina gain of 1 of MΩ 1 MΩ at 10atMHz 10 MHz [22][22] andand references therein), facilitating an easy detection of the bridge’s frequency references therein), facilitating an easy detection of the bridge’s frequency response. response. Figure 1. (Left) Picture of the set-up used to measure the signal coming from the resonator Figure 1. (Left) Picture of the set-up used to measure the signal coming from the resonator (clamped- (clamped-clamped beam) and the movement of the fingertip. (Center) SEM image of the metal clamped beam) and the movement of the fingertip. (Center) SEM image of the metal bridge resonator bridge resonator before releasing process with the different relevant parts highlighted (sample as before received from the releasing foundry). process All the with connections electrical the differenttorelevant parts the CMOS highlighted circuitry (sample asthe are underneath received from the foundry). All the electrical connections to the CMOS circuitry are underneath CMOS protection layer surrounding the resonator area. (Right) Optical image of the released bridge the CMOS protection layer surrounding the resonator area. (Right) Optical image with sketch of the humidity environment change due to breath or finger proximity. of the released bridge with sketch of the humidity environment change due to breath or finger proximity. Sensing is performed through the motional current measurement and then amplified Sensing is performed (transimpedance amplification), through generated thethe between motional bridge current measurement and read-out electrodeand duethen amplified (transimpedance amplification), generated to the bridge movement, which can be approximated using (2): between the bridge and read-out electrode due to the bridge movement, which can be approximated using (2): 𝐼≅𝑉 𝐶 , (2) ∼ ∂C ∂Vac I Vdc + C0 , (2) This assumes that the excitation voltage Vac is=much ∂tsmaller than ∂t the driving voltage Vdc applied to the bridge (capacitive transduction). In (2), the first term corresponds to the This assumes that the excitation voltage Vac is much smaller than the driving voltage motional current, and the second term is the parasitic current; ∂C/∂t is the bridge-readout Vdc applied to the bridge (capacitive transduction). In (2), the first term corresponds to the electrode variable capacitance, and C0 is the static capacitance between the excitation and motional current, and the second term is the parasitic current; ∂C/∂t is the bridge-readout readout electrodes. electrode variable capacitance, and C0 is the static capacitance between the excitation and Motional current measurements were performed using a two-port configuration, i.e., readout electrodes. by applying Vdc to the metal beam and Vac to the excitation electrode and the CMOS am- Motional current measurements were performed using a two-port configuration, i.e., plifier connected to the reading electrode. This configuration minimizes the parasitic cur- by applying Vdc to the metal beam and Vac to the excitation electrode and the CMOS ampli- rent, minimizing the signal level measured out of the resonance, thus enhancing the reso- fier connected to the reading electrode. This configuration minimizes the parasitic current, nance detection (getting a better resonance peak in contrast with the signal out of reso- minimizing the signal level measured out of the resonance, thus enhancing the resonance detection (getting a better resonance peak in contrast with the signal out of resonance), and secondarily, moving the antiresonance peak of the curve (minimum magnitude value peaks shown in Figure 2) away from the resonance (maximum magnitude values shown in Figure 2) curve, allowing for a clear measurement of the frequency of resonance. nance), and secondarily, moving the antiresonance peak of the curve (minimum magni- tude value peaks shown in Figure 2) away from the resonance (maximum magnitude val- ues shown in Figure 2) curve, allowing for a clear measurement of the frequency of reso- Sensors 2023, 23, 8945 4 of 13 nance. Figure 2. Frequency response of metal bridge at different Vdc levels in open-loop and Figure 2. Frequency two-port response of metal bridge at different Vdc levels in open-loop and two-port configuration. configuration. This specific resonator was previously evaluated as a humidity sensor in [15], demon- This specific strating excellentresonator was previously performance with a high evaluated degree of aslinearity a humidity sensor within theinrange [15], demon- of 15% to strating 85% relative humidity (RH). Additionally, it exhibited a very high sensitivityof excellent performance with a high degree of linearity within the range of15% to 3.5 kHz 85% perrelative humidity %RH. Sensor (RH).isAdditionally, linearity it exhibited crucial to accurately a veryrelative measure high sensitivity humidity,of 3.5 kHz a providing per %RH. Sensor linearity straightforward means ofisquantification. crucial to accurately measure relative humidity, providing a straightforward The sensor means limit of quantification. detection is influenced not only by its sensitivity but also by the The sensorand transduction limit detection readout is influenced circuit employed. not only by By itself, theits sensitivity sensitivity of but the also systemby relies the transduction and readout circuit employed. By itself, the sensitivity on the water adsorption capability of the metal used, which is relatively limited. While of the system relies onthis thecharacteristic water adsorption alonecapability may seem of unfavorable, the metal used, which it does notis represent relatively a limited. Whileob- significant this characteristic stacle to achieving alone may seem unfavorable, a high-quality sensor dueittodoes the not represent a significant sub-micrometric dimensions obstacle of the tobridge. achievingThe resonant frequency a high-quality sensor due of the resonator to the is highly responsive sub-micrometric dimensionsto of minute mass the bridge. changes, The resonantsurpassing frequency the ofmoderate the resonatorwaterisvapor highlyadsorption responsivecapability to minuteofmassthe metal layer. changes, The mass the surpassing sensitivity moderate of the watersub-micrometric vapor adsorption clamped-clamped capability of thebeam metalis layer. 4 ag/Hz The(around mass 1.4·103 molecules/Hz). sensitivity This exceptional of the sub-micrometric sensitivitybeam clamped-clamped eliminates the need is 4 ag/Hz (aroundfor an additional 1.4·10 3 mol- sensing layer, ecules/Hz). Thisthereby simplifying exceptional the final sensitivity device. the need for an additional sensing eliminates This CMOS layer, thereby integrated simplifying bridge the final resonator is presented here as a good candidate for device. human This activity CMOS sensing integratedthrough bridge theresonator variationisinpresented ambient humidity, here as awhich goodiscandidate a consequence for of water human evaporation activity sensingproduced through by thefinger proximity. variation in ambient humidity, which is a conse- quence of water evaporation produced by finger proximity. 3. Results In this study, we employed an open-loop configuration with a two-port connection 3. Results to conduct our measurements. The MEMS resonator was connected to a DC voltage In this study, we employed an open-loop configuration with a two-port connection source (Vdc, Keithley 2200-72-1 power supply, Solon, OH, USA), while the actuation to conduct our measurements. The MEMS resonator was connected to a DC voltage source electrode was driven by the AC signal (Agilent network analyzer E5100A, Santa Clara, (Vdc, Keithley 2200-72-1 power supply, Solon, OH, USA), while the actuation electrode CA, USA). The read-out electrode was directly connected to a dedicated CMOS low noise was driven by the AC signal (Agilent network analyzer E5100A, Santa Clara, CA, USA). amplifier, integrated closely with the resonator (not seen in the SEM image of Figure 1). The read-out electrode was directly connected to a dedicated CMOS low noise amplifier, This configuration effectively reduces the parasitic capacitance, thereby improving the integrated closely with the resonator (not seen in the SEM image of Figure 1). This config- detectability of the system, as demonstrated in [23]. To perform the measurements, we uration effectively reduces the parasitic capacitance, thereby improving the detectability utilized a network analyzer (Agilent E5100A), which facilitated accurate data acquisition. Except for the Figure 2 measurements, the sensing measurements along the paper were performed using a Vdc = 30 V and an AC power of −30 dBm (approximately 1 VRMS ). All measurements were performed at ambient conditions (27 ◦ C and 28% RH) and no humidity chamber was used. To begin, the frequency response of the MEMS bridge resonator was thoroughly characterized in terms of the magnitude and phase at various bias voltages, as illustrated Sensors 2023, 23, 8945 5 of 13 in Figure 2. The resonant frequency of the bridge was determined to be around 12.8 MHz for Vdc bias of 30 V, used in the measurements along the paper. In this work, human activity detection is achieved by monitoring the water evaporation from the skin of the human body. When the sensor is placed in proximity to the human body or a specific body part, it can detect changes in the surrounding humidity, effectively serving as an activity detector (see the sketch in Figure 1, Right). The addition of mass to the bridge, resulting from water adsorption, induces a shift in the resonant frequency towards lower frequencies (see Equation (1)), offering various opportunities for human activity detection. Specifically, we measured the phase shift and magnitude shift in the frequency response of the MEMS resonator to detect finger movements. Schematically, the measurement procedure follows the steps listed below: (a) First, we have selected a specific point on the magnitude or phase curves correspond- ing to the resonant frequency of the bare bridge (without added water, and with the sensor far from human body). For the magnitude curve, the point is at the maximum value of the curve. For the phase curve, this point is at the maximum slope of the phase curve (see Figure 2). (b) After that, the measurements are conducted with a frequency sweep span of zero. By fixing the frequency and recording the phase or magnitude values at the chosen point, a straight (horizontal) line can be obtained on the magnitude/phase versus the measurement event plot (see Figure 3, baseline, as an example). This line will maintain a constant measured value until a mass addition event occurs. (c) When we approximate a human body part, in our case the fingertip, or breathing over the sensor, the frequency response curve shifts to the left (towards lower frequencies) due to the added mass (water) and therefore, the measurement (with a frequency sweep span of zero) reflects this change by altering the constant value (magnitude or phase) to a lower value (see Figure 3, wet foam, as an example). When the fingertip is Sensors 2023, 23, x FOR PEER REVIEW moved far away or we stop breathing onto the bridge, the measurement returns 6 of 14to the initial value, showing a dewetting process of the metal bridge. Figure 3. Magnitude shift at a fixed frequency (resonance frequency) bringing foam closer and further Figure 3. Magnitude shift at a fixed frequency (resonance frequency) bringing foam closer and fur- away from the metal bridge. Results for dry and wet foam. ther away from the metal bridge. Results for dry and wet foam. To obtain the plots of the different figures we have corrected the values obtained using theTo obtainsignal baseline the plots levelofinthe different order to showfigures only we the have corrected relative the values or shifts (magnitude obtained phase) us- with ing the baseline signal level in order to show only the relative shifts (magnitude respect to the baseline. It is important to point out that measurements can be performed or phase) with respect using to the the phase orbaseline. It is important the magnitude to point curves; there is noout thatreason special measurements to choosecan be per- a magnitude formed using the phase or the magnitude curves; there is no special reason to choose a magnitude or phase measurement, except, as we will see in the following sections, if high sensitivity is required, in which case it is better to choose a phase instead of magnitude measurement. Sensors 2023, 23, 8945 6 of 13 or phase measurement, except, as we will see in the following sections, if high sensitivity is required, in which case it is better to choose a phase instead of magnitude measurement. 3.1. Foam Measurements To ascertain whether the response of the microbridge sensor is indeed attributed to hu- midity (water addition to the bridge) or another phenomenon (for example, a temperature change), we conducted preliminary measurements. Initially, we selected the maximum point on the magnitude curve and fixed the frequency to perform a zero-span measurement. To simulate different scenarios, we employed a foam material and repetitively moved it closer to and further away from the sensor (using a manual stage). Using foam material is Sensors 2023, 23, x FOR PEER REVIEW advantageous because, in addition to accumulating water in its porous structure, it does not generate heat as the human body does, which enables us to separate the effects of humidity on the metal bridge from the thermal effects. In Figure with a manual 3, two distinctmicropositioning measurements using system. Theare foam measurements presented. were taken The first at distances mea- ing from 10 mm to 3.5 mm (see the schematic in surement involved dry foam, where the magnitude value exhibited only slight changesFigure 4a). To obtain each phas compared to thepoint, we computed baseline. Conversely,the theaverage value over time second measurement (14 s),wet utilized which foam,isresulting shown in the Fi in a significant inset graphs; alteration themagnitude of the statistical error valueiswhen presented as error the foam was bars. brought closer to the To facilitatedown sensor (from a few centimeters the measurements of our from to a few millimeters CMOS thesystems, sensor). our chip These was wire-bond findings indicate that thea home-made printed circuit changes in magnitude board correlated are indeed with SMA with connectors. The bondings the presence of water, employe confirming thenon-electrically sensitivity of theprotected sensor toaluminum connections. Due to the limitations of this setu humidity variations. had to limit our measurements to a minimum distance of 3.5 mm over the water 3.2. Sensor Calibration tank to(Phase ensure Shift theversus Distance) integrity of the system and avoid the contact of the water and bondings. Once we have The results stablished that the ofphase these measurements (or magnitude)are depicted shift in Figure measurement 4. is useful to detect humidity Fromchanges,the we proceeded observed data,toit calibrate is evident the response that as the of the microbridge distance decreases from 6 m based on the distance smallerfrom a water distances, source. there In order to change is a significant accomplish this, we in phase. conducted As the system mea-moves closer surements by fixing the frequency at the resonance value of the resonator and, water tank, the phase value loses stability, exhibiting values, in the time domainmeasuring the phase shift,smoothly varying the varydistance between form in a non-regular our system insteadand of ina awater-filled straight linetank (seewith Figurea 4 insets manual micropositioning behaviour system. The measurements can be attributed were taken to the dynamic at distances equilibrium ranging of water from and d absorption 10 mm to 3.5 mm tion(see ontothe theschematic in Figure bridge, resulting in 4a). To obtain fluctuations ineach phasemeasurement. the phase shift point, we computed the average value over time (14 s), which is shown in the Figure 4 inset graphs; the statistical error is presented as error bars. Figure 4. PhaseFigure shift versus chip-to-water 4. Phase tank distance,tank shift versus chip-to-water measured at measured distance, the resonant frequency. at the resonant frequen (a) Set-up used to perform Set-up usedthe measurement. to perform (b) Phase shift the measurement. versusshift (b) Phase distance plot. versus In theplot. distance insets In the inset of (b) oscillationsoscillations over time ofover thetime phaseof measurement the phase measurement are shown.are shown. To facilitate theItmeasurements is worth notingofthat ourthe CMOS systems, top metal used our in chip was wire-bonded the bridge constructionon does not p a home-made printed specific circuit board with properties. water-absorption SMA connectors. The bondings The presence employed of a dynamic usesheet ov water non-electricallybridge protected aluminum leads connections. to a continuous Due to formation andtheevaporation limitations of of this the setup, water we layer. This had to limit ourmay measurements to a minimum distance of 3.5 mm over the water-filled tank not be critical for human activity detection, such as the finger movement det proposed in this work. 3.3. Application: Finger Movement In this section, we discuss the experiments that were performed to prove the of our system to detect finger movements. We use both the phase and the magnitud Sensors 2023, 23, 8945 7 of 13 to ensure the integrity of the system and avoid the contact of the water and wire-bondings. The results of these measurements are depicted in Figure 4. From the observed data, it is evident that as the distance decreases from 6 mm to smaller distances, there is a significant change in phase. As the system moves closer to the water tank, the phase value loses stability, exhibiting values, in the time domain, that smoothly vary in a non-regular form instead of in a straight line (see Figure 4 insets). This behaviour can be attributed to the dynamic equilibrium of water absorption and desorption onto the bridge, resulting in fluctuations in the phase measurement. It is worth noting that the top metal used in the bridge construction does not possess specific water-absorption properties. The presence of a dynamic water sheet over the bridge leads to a continuous formation and evaporation of the water layer. This aspect may not be critical for human activity detection, such as the finger movement detection proposed in this work. 3.3. Application: Finger Movement In this section, we discuss the experiments that were performed to prove the ability of our system to detect finger movements. We use both the phase and the magnitude shift as measurement variables. 3.3.1. Phase Shift Detection Sensors 2023, 23, x FOR PEER REVIEW 8 of 14 Figure 5 illustrates the detection of finger movement using phase as a measurement pa- rameter. In these measurements, the frequency was fixed at the resonance point, specifically at the maximum slope on the phase curve. During the experiment, the finger was repeatedly centimeters. moved closer Additionally, while to and further maintaining away a constant distance from the microbridge sensor in the z-direction within (per- a few centimeters. pendicular to the sensor plane), the finger was moved left and right (5 cm of Additionally, while maintaining a constant distance in the z-direction (perpendicular to thelateral travel). The minimum sensor plane),distance between the finger the finger was moved leftand andsensor was right (5 cmfixed to fewtravel). of lateral millimeters (taking The minimum into account a variability of the distance due to non-precise finger movements) distance between the finger and sensor was fixed to few millimeters (taking into account a to preserve chip-to-board variability ofbondings. the distanceRemarkably, both types due to non-precise of movements finger movements)resulted in achip-to-board to preserve similar re- sponse from the bridge, allowing us to state that the resonator has bondings. Remarkably, both types of movements resulted in a similar response from the an omnidirectional response bridge, that is only allowing usdependent to state thatonthe theresonator radial distance to the bridge. response that is only has an omnidirectional dependent on the radial distance to the bridge. (a) (b) Figure 5. Finger movement detection measuring phase shift. (a) Up and down movement; (b) left Figure 5. Finger and right movement detection measuring phase shift. (a) Up and down movement; (b) left movement. and right movement. As the humidity level increases, the MEMS resonator experiences a reduction in its As the humidity resonance level frequency, increases, resulting in athe MEMSin decrease resonator experiences phase (negative a reduction phase in its shift) when it is resonance frequency, resulting in a decrease in phase (negative phase shift) when it is measured at the dry device resonant frequency. To detect the resonant frequency variations, measured at the dry device resonant frequency. To detect the resonant frequency varia- tions, and therefore the %RH, we can utilize the slope of the frequency response curve. In Figure 2, at a DC voltage of 30 V, the slope of the phase curve (at the resonant point) is approximately −5.7·10−4 deg/Hz. When the finger is moved closer to the detector, reaching a distance of a few milli- Sensors 2023, 23, 8945 8 of 13 and therefore the %RH, we can utilize the slope of the frequency response curve. In Figure 2, at a DC voltage of 30 V, the slope of the phase curve (at the resonant point) is approximately −5.7·10−4 deg/Hz. When the finger is moved closer to the detector, reaching a distance of a few mil- Sensors 2023, 23, x FOR PEER REVIEWlimeters above it, the phase experiences a decrease of approximately 8 degrees (in this 9 o example). This decrease corresponds to an approximate frequency shift of 14 kHz, a variation of 4 in %RH, indicating a mass addition of around 56 femtograms (equivalent to approximately 1.7·109 molecules of water), based on the calibration conducted in [15] that 10−11 g·cm (3.4·results −2 ·Hz from −1 ). Proportionally moving the fingertip upresonant to the and down ∆f/f0 ,the (bending frequency, thisfinger). We also me shift amounts to around 1100 ppm. ured the g-force-related parameters. For these measurements, we selected a fixed point Conversely, the phase curve when the finger isto corresponding moved further away, the resonance the humidity frequency of the decreases, bridge.and the phase value rapidly recovers. This phenomenon demonstrates the efficient water desorption Figures 6 and 7 illustrate comparisons that were observed between the microme capability of the metal bridge, which corresponds to its low humidity hysteresis cycle. bridge Theresponse and phase shift theobserved slope measurements in Figureobtained from the22.2 5 is approximately commercial deg/s, which gyroscope. trans- Th islates a strong correlation between the phase changes in the bridge measurement to 39 kHz/s. This remarkable slope enables the sensor to achieve a high sensitivity and the c responding gyroscope responses. and rapid response. In addition to the fingercompares Figure 6 specifically movementthe measurements described bridge response earlier, with we conducted the angular velocity of further experiments using a commercial gyroscope (LSM6DSL from STmicroelectronics, fingertip. It is evident that the maximum phase shift points in the microbridge respo Geneva, Switzerland) with a resolution of 6.11·10−4 rad/s and a gravity sensor (Samsung coincides with a point situated between the consecutive minimum and maximum angu Gravity Sensor, Suwon, Republic of Korea) with a resolution of 0.001 g. We utilized the velocity gyroscope’svalues. When to capabilities the fingertip measure approaches angular movements,thespecifically sensor, the theangular angular velocity velocity reache maximum that results from moving the fingertip up and down (bending the finger). We also measuredbridge, negative value. As the fingertip movement is halted near the sensor angular velocity value the g-force-related becomes parameters. For zero. At this point, the these measurements, microbridge we selected a fixedresponse point on thepresents phase curve corresponding to the resonance frequency of the minimum phase measurement value (maximum phase shift), corresponding with bridge. pointFigures where 6theand 7 illustrate comparisons that were observed between the micrometric finger is closer by. Subsequently, as the fingertip is retracted, the angu bridge response and the measurements obtained from the commercial gyroscope. There velocity is a strongreaches a maximum correlation between the positive value, and phase changes in thethebridge phase measurement measurement and returns the to original value.gyroscope responses. corresponding (a) (b) Figure 6. Fingertip movement (up and down) sensed by a commercial gyroscope measuring the Figure 6. Fingertip movement (up and down) sensed by a commercial gyroscope measuring angular velocity (a) in comparison with the detection by the microbridge sensor (phase shift) (b). angular velocity (a) in comparison with the detection by the microbridge sensor (phase shift) (b Sensors 2023, 23, x FOR PEER REVIEW 10 of 14 Sensors 2023, 23, 8945 9 of 13 (a) (b) Fingertip movement Figure 7. Fingertip movement(up (upandanddown) down)sensed sensedbybya acommercial commercial accelerometer accelerometer measuring measuring g- force (a)(a) g-force (axis perpendicular (axis perpendicularto to earth gravity earth direction) gravity in comparison direction) with in comparison the the with detection by the detection by mi- the crobridge sensor microbridge (phase sensor shift) (phase (b).(b). shift) This observation Figure highlights 6 specifically compares thethe close relationship bridge responsebetween with the the phase velocity angular measurementsof the fingertip. It is evident of the microbridge andthat the maximum the angular velocityphase of theshift pointsasincaptured fingertip the microbridge response by the gyroscope. coincides withnotable Another a point observation situated between the consecutive from Figure 6 is that the minimum maximum and and maximum minimum angular val- velocity ues of thevalues. angularWhen the fingertip velocity approaches tend to remain the sensor, relatively the angular consistent. However, velocity this is reaches not the acase maximum negative for the phase value. As the measurements fingertip obtained frommovement the bridge is response. halted near Thethe sensor bridge bridge, response the angular is highly velocity value dependent on thebecomes proximityzero. of At thethis point, the fingertip, andmicrobridge its sensitivityresponse increasespresents expo- the minimum nentially with phase measurement decreasing distance. value (maximum As a result, even when phasetheshift), corresponding fingertip is moved at with the a max- point where the finger is closer by. Subsequently, as the fingertip imum angular velocity, deep peaks in the bridge measurements cannot be achieved unless is retracted, the angular velocity the fingerreaches a maximum is brought positive value, in close proximity to the and the phasewithin microbridge, measurement returns to its a few millimeters. original value. This fact highlights the crucial role of distance in influencing the sensitivity and mag- This nitude of observation highlightsSignificant the bridge response. the close relationship changes in the between bridgethe phase measurements measurements can be ob- of the microbridge tained only whenand thethe angular finger velocity is brought of to close thethe fingertip as captured microbridge, whereby thethe gyroscope. proximity effect Another becomes morenotable observation from Figure 6 is that the maximum and minimum values pronounced. of theFigure angular 7 presentstend velocity the to remain relatively acceleration (g-force)consistent. measurement,However,whichthiscomplements is not the case thefor the phase measurements obtained from the bridge response. The bridge angular velocity measurement discussed earlier. In this case, we focused on the y-direc- response is highly dependent tion g-forceon the proximity component, of the fingertip, as specified by the and its sensitivity commercial increases sensor’s exponentially capabilities. with In the y-axis decreasing distance. As a result, even when the fingertip is moved direction, which represents one of the in-plane directions perpendicular to the Earth’s at a maximum angular velocity, deep gravity, the peaksvalue g-force in theisbridge close tomeasurements zero when thecannot fingertipbe is achieved unless in a neutral the finger position. How- is brought in close proximity to the microbridge, within a few millimeters. ever, as the finger is tilted, the g-force value deviates from zero and reaches a maximum This fact highlights the crucial role of distance in influencing the sensitivity and negative value at the maximum tilt angle. magnitude of the bridge response. Significant changes in the bridge measurements can be To provide some context, a g-force value of 0.5 corresponds to a fingertip tilt angle of obtained only when the finger is brought close to the microbridge, where the proximity approximately 45 degrees. As the fingertip is retracted and returns to a position perpen- effect becomes more pronounced. dicular to the Earth’s gravity, the g-force value returns to zero. Just before initiating the Figure 7 presents the acceleration (g-force) measurement, which complements the finger retraction, when the fingertip is closest to the microbridge, we observe the mini- angular velocity measurement discussed earlier. In this case, we focused on the y-direction mum value of the phase measurement at our sensor. Consequently, as the g-force value g-force component, as specified by the commercial sensor’s capabilities. In the y-axis for the y-direction returns to zero, the phase measurement also returns to its initial value. direction, which represents one of the in-plane directions perpendicular to the Earth’s This relationship between the g-force measurements in the y-direction and the corre- gravity, the g-force value is close to zero when the fingertip is in a neutral position. However, sponding phase measurements further emphasizes the interconnected nature of the mi- as the finger is tilted, the g-force value deviates from zero and reaches a maximum negative crobridge response and the finger movement. value at the maximum tilt angle. To provide some context, a g-force value of 0.5 corresponds to a fingertip tilt angle of 3.3.2. Finger Movement: Magnitude Shift Detection approximately 45 degrees. As the fingertip is retracted and returns to a position perpen- dicular to the Earth’s gravity, the g-force value returns to zero. Just before initiating the To measure the magnitude shift, we utilized the maximum point on the magnitude curve as the fixed frequency measurement corresponding to the device resonant fre- quency. When mass is added to the bridge, a decrease in the magnitude at this specific Sensors 2023, 23, 8945 frequency point can be observed, as depicted in Figure 8. 10 of 13 The slope of the magnitude curve after the resonance point is approximately −1.2 × 10−4 dB/Hz. As the finger is moved closer to the sensor by a few millimeters, the magnitude decreases by approximately finger retraction, when the 0.15 dB. This fingertip change is closest is equivalent to the microbridge,to awe frequency shift observe the of ap- minimum proximately 1.25 kHz. In terms of proportionality to the resonance frequency, value of the phase measurement at our sensor. Consequently, as the g-force value for this shiftthe amounts to around y-direction 100 returns to ppm, zero, indicating a slightly lower the phase measurement sensitivity also returns tocompared its initialto the phase value. detectionThismethod, as expected: relationship betweenatthe theg-force resonant frequency point, measurements in thethe slope of the y-direction andphase the cor- curve is maximum; responding phasehowever, this is not measurements the case further for the magnitude emphasizes curve, as itnature the interconnected is already of the close to its maximum microbridge point response andand, therefore, the finger exhibits a relatively small slope. movement. Similarly, when the finger is moved further away, the magnitude recovers rapidly, as does Finger 3.3.2. Movement: the phase Magnitude detection method. Shift Detection The slope of the magnitude recovery is approxi- mately To 0.8measure dB/s, corresponding the magnitude to ashift, frequency shift of we utilized approximately the maximum point 6.6 kHz on thepermagnitude second. curve as the fixed frequency measurement corresponding to the device resonant frequency. When mass is added to the bridge, a decrease in the magnitude at this specific frequency point can be observed, as depicted in Figure 8. Figure 8. Fingertip movement (up and down) detection using magnitude shift. Figure 8. Fingertip movement (up and down) detection using magnitude shift. The slope of the magnitude curve after the resonance point is approximately −1.2 × 10−4 dB/Hz. As the finger is moved closer to the sensor by a few millimeters, the 3.4. Breath Detection magnitude decreases by approximately 0.15 dB. This change is equivalent to a frequency The shift of last experiment 1.25 approximately that kHz. we performed In terms ofisproportionality breathing 10 cmtofrom the sensorfrequency, the resonance with a regular (Figure 9) and irregular (Figure 10) respiration rate. For this experiment this shift amounts to around 100 ppm, indicating a slightly lower sensitivity compared we used to phase detection, fixing the measurement of the phase at the resonant frequency the phase detection method, as expected: at the resonant frequency point, the slope of point. Wethe can see incurve phase Figureis 9maximum; that, whenhowever, we exhalethis air is over notthe thesensor, case forthe phase the decreases magnitude abruptly curve, as it is due to the close already shift to to low frequencies, its maximum which point and,istherefore, a result ofexhibits the increase in humidity. a relatively small After slope.each exhalation, the phase Similarly, whenrecovery the fingeris is not complete. moved After further thethe away, lastmagnitude exhalation,recovers the phase takesas rapidly, roughly does the four seconds phase to abruptly detection method. recover the initial The slope of thevalue (the lastrecovery magnitude jump). is approximately 0.8 dB/s, corresponding to a frequency shift of approximately 6.6 kHz per second. 3.4. Breath Detection The last experiment that we performed is breathing 10 cm from the sensor with a regular (Figure 9) and irregular (Figure 10) respiration rate. For this experiment we used phase detection, fixing the measurement of the phase at the resonant frequency point. We can see in Figure 9 that, when we exhale air over the sensor, the phase decreases abruptly due to the shift to low frequencies, which is a result of the increase in humidity. After each exhalation, the phase recovery is not complete. After the last exhalation, the phase takes roughly four seconds to abruptly recover the initial value (the last jump). Sensors 2023, 23, x FOR PEER REVIEW 12 of 14 Sensors 2023, 23, x FOR PEER REVIEW 12 of 14 Sensors 2023, 23, 8945 11 of 13 Figure 9. Phase change during six seconds of regular breathing. Figure 9. Figure Phasechange 9. Phase changeduring duringsix sixseconds seconds of of regular regular breathing. breathing. Figure 10. Figure Phase change 10. Phase change during during nine nine seconds seconds of of irregular irregular breathing. Figure 10. Phase change during nine seconds of irregular breathing. This abrupt This abrupt recovery recovery of of the the initial initial value value is is now now under under discussion, discussion, butbut it it seems seems thatthat we This can abrupt have a recovery water layer of on the the initial bridge value which is now under evaporates discussion, progressively we can have a water layer on the bridge which evaporates progressively until a system of but it until aseems system that of water we can islands have a is formed. water layer The on water the islands bridge increase which the evaporates evaporation velocity progressively water islands is formed. The water islands increase the evaporation velocity and, together untiland, a together system of with the water with the fact fact that islands that the themass is formed. The mass remaining ononthe water islands remaining bridge increase the bridge in in the this situation evaporation this is too velocity situation small, is too and, causes small, togetherthe causes final with phase the factjump. that Figure the mass10 shows remainingan irregular on the breath bridge pattern, in this demonstrating situation the final phase jump. Figure 10 shows an irregular breath pattern, demonstrating the abil- is too the small, ability causes of the the sensor final to clearly follow the breath rates. ity of thephase sensorjump. Figure to clearly 10 shows follow an irregular the breath rates. breath pattern, demonstrating the abil- ity of the sensor to clearly follow the breath rates. 4. Conclusions 4. Conclusions In this study, we have introduced a simple sub-micrometric metal bridge integrated 4. Conclusions withInIn CMOS this study, technologywe have introduced as aintroduced a simple sub-micrometric promising candidate for a fast and highly metal bridge integrated sensitive with this study, CMOS we have technology as a promising acandidate simple sub-micrometric for a fast and metal highly bridgenon-contact sensitive integrated non-con- human with CMOS activity sensor. The technology as a sensor promisingleverages its ability candidate for a to detect fast and humidity highly resulting sensitive from non-con- tact waterhuman activity evaporation sensor. on the The body’s sensor leverages surface. Both its phaseability and to detect magnitude humidity changes resulting can be tact from human water for activity sensor. evaporation on theThebody’s sensorsurface. leveragesBoth itsphase ability to magnitude and detect humidity changes resulting can be monitored from water for this evaporationpurpose. on the body’s surface. Both phase and magnitude changes can be monitored this purpose. monitored for this purpose. Sensors 2023, 23, 8945 12 of 13 In the case of phase change, moving the finger closer or further away from the sensor induces a frequency shift of approximately 14 kHz (as shown in Figure 5). This frequency shift corresponds to a relative humidity modification of around 4%, indicating a high sensitivity of the sensor. Taking into account the frequency of resonance and the calibration performed in [15], our sensor presents a sensitivity of 0.03% of variation of frequency per %RH, an average response time of 1 s, and 2 s of average recovery time. Although we conducted the experiments using a network analyzer in an open-loop mode, it is easily adaptable to a closed-loop mode using a frequency counter. Despite the small quantity of water adsorbed by the metal, the high resonant frequency, good mass sensitivity, and readout circuitry enable excellent human activity detection. Additionally, the limited amount of water adsorbed facilitates a quick recovery of the phase or magni- tude to their previous values, enabling rapid movement detection. Good correspondence between our sensor for finger movement detection with a commercial gyroscope has been demonstrated with the advantage of providing a non-contact measurement system. Other authors referenced here have presented sensors with high sensitivity. As far as we know, sensors based on surface acoustic wave resonators are the best. For example, the authors of [10] have obtained a sensitivity of 0.005% of variation of frequency of resonance per %RH. The authors of [17] have proposed a 4-anchored plate oscillator with a variation of 0.02% of frequency of resonance per %RH. These two examples have a better sensitivity than our sensor, but a complex fabrication is needed, specifically for SAW sensors. Further, technical uncertainties can appear when using the proposal outlined in [17], due to the use of distributed holes in the plate to accomplish resonator releasing. In [18], the authors have obtained a sensor comparable with our proposal in response and recovery times. For the response time, the authors of [18] have obtained 1 s as in our case. For recovery time, the authors have obtained 2.8 s, which is very close to the 2 s of our proposal. However, in [18], a film of MoS2 is needed, whereas in our proposal no absorbing layer is needed. In comparison with [17–20], our metal bridge MEMS resonator has a good sensitivity in the minimum detecting area, even without the use of a specific sensing layer over the resonator. Overall, this integrated metal bridge sensor demonstrates the potential to serve as a highly sensitive and fast human activity sensor, making it a valuable tool for various applications. Author Contributions: Conceptualization, F.T. and N.B.; methodology, F.T. and N.B.; formal analysis, F.T.; investigation, F.T.; data curation, F.T.; writing—original draft preparation, F.T.; writing—review N.B. and A.U.; supervision, N.B.; project administration, N.B. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Spanish Ministry of Science and Innovation (Spanish Government) with project number: PID2019-108270RB-I00. Data Availability Statement: Data supporting figures within this paper are available from the corresponding author upon reasonable request. Conflicts of Interest: The authors declare no conflict of interest. References 1. Hernández, A.; Morales, V.; García, V. Finger tapping test. Precisión del diseño de medidas entre muestras de deportistas de élite y no deportistas. Cuad. De Psicol. Del Deporte 2011, 11, 29–43. [CrossRef] 2. Memon, M.M.; Liu, Q.; Manthar, A.; Wang, T.; Zhang, W. Surface acoustic wave humidity sensor: A review. Micromachines 2023, 14, 945. [CrossRef] [PubMed] 3. Tai, H.L.; Wang, S.; Duan, Z.H.; Jiang, Y.D. Evolution of breath analysis based on humidity and gas sensors: Potential and challenges. Sens. Actuators B Chem. 2020, 318, 128104. [CrossRef] 4. Dai, J.; Zhao, H.; Lin, X.; Liu, S.; Liu, Y.; Liu, X.; Fei, T.; Zhang, T. Ultrafast Response Polyelectrolyte Humidity Sensor for Respiration Monitoring. ACS Appl. Mater. Interfaces 2019, 11, 6483–6490. [CrossRef] [PubMed] Sensors 2023, 23, 8945 13 of 13 5. Liu, Y.; Chen, L.; Zhang, Q.; Huang, J. A breathable capacitive humidity sensor based on electrospun PVDF/GO composites. In Proceedings of the Transducers Conference, Kyoto, Japan, 25–29 June 2023; Institute of Electrical Engineers of Japan: Tokyo, Japan, 2023; pp. 1081–1084, ISBN 978-4-88686-435-2. 6. Su, Y.; Li, C.; Li, M.; Li, H.; Xu, S.; Qian, L.; Yang, B. Surface acoustic wave humidity sensor based on three-dimensional architecture graphene/PVA/SiO2 and its application for respiration monitoring. Sens. Actuators B Chem. 2020, 308, 127693. [CrossRef] 7. Jang, I.R.; Jung, S.I.; Ryu, C.; Park, J.; Padhan, A.; Yu, J.; Keum, H.; Kim, H.J. Graphene oxide integrated surface acoustic wave humidity sensor with simultaneous multi-frequency operation. In Proceedings of the Transducers Conference, Kyoto, Japan, 25–29 June 2023; Institute of Electrical Engineers of Japan: Tokyo, Japan, 2023; pp. 1833–1836, ISBN 978-4-88686-435-2. 8. Farahani, H.; Wagiran, R.; Hamidon, M.N. Humidity sensors principle, mechanism, and fabrication technologies: A Comprehen- sive Review. Sensors 2014, 14, 7881–7939. [CrossRef] 9. Niu, G.; Wang, Z.; Xue, Y.; Yan, J.; Dutta, A.; Chen, X.; Wang, Y.; Liu, C.; Du, S.; Guo, L.; et al. Pencil-on-paper humidity sensor treated with NaCl solution for health monitoring and skin characterization. Nano Lett. 2023, 23, 1252–1260. [CrossRef] 10. Dong, H.; Li, D.; Pang, J.; Zhang, Q.; Xie, J. Highly sensitive and fast-response humidity sensor based on SAW resonator and MoS2 for human activity detection. In Proceedings of the IEEE 34th International Conference on Micro Electro Mechanical Systems (MEMS), Gainesville, FL, USA, 25–29 January 2021; pp. 322–325. [CrossRef] 11. Jung, S.I.; Jang, I.R.; Ryu, C.; Park, J.; Padhan, A.M.; Kim, H.J. Graphene oxide decorated multi-frequency surface acoustic wave humidity sensor for hygienic applications. Sci. Rep. 2023, 12, 6838-1–6838-11. [CrossRef] 12. Chen, G.; Dai, X.; Liu, Y.; Mao, H.; Chen, D. A highly sensitive humidity sensor based on nanoforests. In Proceedings of the IEEE 34th International Conference on Micro Electro Mechanical Systems (MEMS), Gainesville, FL, USA, 25–29 January 2021; pp. 330–333. [CrossRef] 13. Xu, J.; Peiner, E. Micromachined silicon cantilever resonator-based humidity sensors for multifunctional applications. In Proceedings of the IEEE 34th International Conference on Micro Electro Mechanical Systems (MEMS), Gainesville, FL, USA, 25–29 January 2021; pp. 326–329. [CrossRef] 14. Arcamone, J.; Sansa, M.; Verd, J.; Uranga, A.; Abadal, G.; Barniol, N.; van den Boogaart, M.; Brugger, J.; Pérez-Murano, F. Nanomechanical mass sensor for spatially resolved ultrasensitive monitoring of deposition rates in stencil lithography. Small 2009, 5, 176–180. [CrossRef] [PubMed] 15. Verd, J.; Sansa, M.; Uranga, A.; Pérez-Murano, F.; Segura, J.; Barniol, N. Metal microelectromechanical oscillator exhibiting ultra-high water vapor resolution. Lab Chip 2011, 11, 2670–2672. [CrossRef] [PubMed] 16. Perelló-Roig, R.; Verd, J.; Bota, S.; Soberats, B.; Costa, A.; Segura, J. CMOS-MEMS VOC sensors functionalized via inkjet polymer deposition for high-sensitivity acetone detection. Lab Chip 2021, 21, 3307–3315. [CrossRef] [PubMed] 17. Perelló-Roig, R.; Orvay, F.; Paúl, I.; Verd, J.; Bota, S.; Segura, J. Fully integrated front-end CMOS-MEMS transducer for low-cost real-time breath monitoring. In Proceedings of the IEEE Sensors Conference, Dallas, TX, USA, 30 October–2 November 2022. [CrossRef] 18. Li, D.; Liu, W.; Zhu, B.; Qu, M.; Zhang, Q.; Fu, Y.; Xie, J. Machine learning-assisted multifunctional environmental sensing based on a piezoelectric cantilever. ACS Sens. 2022, 7, 2767–2777. [CrossRef] [PubMed] 19. Nuryadi, R.; Djajadi, A.; Adiel, R.; Aprilina, L.; Aisah, N. Resonance frequency change in microcantilever-based sensor due to humidity variation. Mater. Sci. Forum 2013, 737, 176–182. [CrossRef] 20. Dennis, J.; Ahmed, A.; Khir, M. Fabrication and characterization of a CMOS-MEMS humidity sensor. Sensors 2015, 15, 16674–16687. [CrossRef] [PubMed] 21. Shmid, S.; Villanueva, L.G.; Roukes, M.L. Fundamentals of Nanomechanical Resonators, 1st ed.; Springer International Publishing: Cham, Switzerland, 2016; p. 39. [CrossRef] 22. Verd, J.; Uranga, A.; Abadal, G.; Teva, J.L.; Torres, F.; López, J.; Pérez-Murano, F.; Esteve, J.; Barniol, N. Monolithic CMOS MEMS oscillator circuit for sensing in the attogram range. IEEE Electron. Dev. Lett. 2008, 2, 146–148. [CrossRef] 23. Verd, J.; Uranga, A.; Teva, J.; López, J.L.; Torres, F.; Esteve, J.; Abadal, G.; Pérez-Murano, F.; Barniol, N. Integrated CMOS-MEMS with on-chip readout electronics for high-frequency applications. IEEE Electron. Dev. Lett. 2006, 27, 495–497. [CrossRef] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.