ES2245614B1 - Read system for monolithic semiconductor resonant mechanical transducer element, has polarizer which is connected between input and output nodes of amplifier in order to polarize input node by connecting same to output node - Google Patents

Abstract

The read system comprises a voltage amplifier which defines input and output nodes. The output of the transducer is connected to the input node. The amplifier amplifies the voltage at the input node which corresponds to the voltage at the output of the transducer. A polarizer is connected between the input and output nodes of the amplifier in order to polarize the input node by connecting same to the output node.

Description

Reading system for a transducer element mechanic.

The present invention relates to a system of reading for a mechanical transducer element, and more specifically, to a very high sensitivity reading system for resonators monolithic mechanics

Background of the invention

At present there is great interest in the development of sensor reading systems, of very high sensitivity. Examples of microelectromechanical systems ( MEMS ) comprise sensors to detect different types of physical or chemical properties [O. Brand, H.Baltes. Micromachined resonant sensors. An overview Sensors, vol.4, Wiley-VCH (1998)] . "Physical" MEMS sensors are based on an element that transduces the physical magnitude to be detected (eg mass, acceleration, pressure, electrical charge, etc.) in an electrical signal. In "chemical" or "biochemical" sensors, the compound or molecule to be detected is selectively absorbed by the transducer element, thanks to the previous functionalization of its surface by a layer of a material that "traps" the molecule to be measured. When this entrapment occurs, the mass of the molecule that is deposited is detected by the sensor.

A further reduction in the dimensions of the mechanical transducer element leads to a new generation of systems called nanoelectromechanics ( NEMS ), which represent an improvement in sensitivity, spatial resolution, energy efficiency and temporal response. Such NEMS systems are described, for example, in [MLRoukes. Nanoelectromechanical systems. Technical Digest of the 2000 Solid-State Sensor and Actuator Workshop, Hilton Head Isi., SC, 6 / 4-8 / 2000 (ISBN 0-9640024-3-4)] , [HG Craighead, Nanoelectromechanical Systems. Science, vol. 290, 1532-1535 (2000)] or [R. Berger, Ch. Gerber, HP Lang, JKGimzewski. Micromechanics: A Toolbox for Femtoscale Science: "Towards a Laboratory on a Tip". Microelectronic Eng. 35, 373-379 (1997)] .

MEMS / NEMS systems based on levers (see Fig. 1), consisting of a lever or several matrix-shaped levers, are characterized by the great simplicity of the transducer element [NVLavrik, MJSepaniak, PGDatskos. Cantilever transducers as a platform for chemical and biological sensors. Review of Scientific Instruments, vol. 75, No. 7, 2229-2253 (2004)] . Its operation is based on the transduction that produces the magnitude that you want to measure in the mechanical performance of the lever of micrometric and sub-micrometric dimensions. The principle of transduction consists in the deflection of the lever (DC or continuous deflection sensors) or in the alteration of some characteristic of its vibratory movement, produced in this case by an excitation system (AC or resonant sensors). The deflection or variation of the oscillation of the lever is detected by an optical or electrical reading system that provides an electrical signal at the sensor output.

There are several advantages of using said lever structure as a sensor, among which the simplicity of the transducer element, the possibility of array configuration ( arrays ), and the transduction capacity of almost any signal domain with great sensitivity. Although the sensors based on levers or micro-levers have their origin and main impulse in atomic probe microscopes (mainly force microscopes), these sensors allow to detect a multitude of physical and / or chemical properties with high resolution, such as temperature, pressure, acceleration , mass, or chemical reactions that produce changes in surface stress [HPLang, M.Hegner, E.Meyer, Ch.Gerber. Nanomechanics from atomic resolution to molecular recognition 30 based on atomic force microscopy technology. Tutorial. Nanotechnology, 13.R29-R36 (2002)] .

The methods of excitation and detection of the oscillation of the lever can be optical, piezoresistive or piezoelectric, capacitive, magnetic or temperature dependent. Currently, the optical detection systems (laser diode and photodetector) are the ones that can be translated into a higher resolution due to the electrical decoupling between the mechanical and the electronic system, as described in [FMBattiston, JPRamseyer, HPLang, MKBalter, Ch. Gerber, JK G imzewski, E. Meyer, HJ Güntherodt. A chemical sensor based on a microfabricated cantilever array with simultaneous resonance-frequency and bending readout. Sensors and Actuators (B), 77, 122-131 (2001)] and [T.Ono, X.Li, H.Miyashita, M.Esashi. Mass sensing of adsorbed molecules in sub-picogram sample with ultrathin silicon resonator. King. Sci.lnstrum. 74. 1240-1243. (2003)] . But this type of detection cannot be applicable in the case of levers of nanometric dimensions (below 100 nm), such as NEMS systems, due to the diffraction limit, in addition to assuming greater complexity in the system and a lower degree of integrability These drawbacks are mainly based on the fact that a laser beam that focuses on the lever and a photodetector that can detect it is necessary, so that, depending on the shadow made by the laser, a signal is obtained. Therefore, the complexity in the realization of integrated systems of this type is evident, since it is necessary to mechanically align the laser beam with the mechanical part (eg a lever-based sensor) and the photodetector.

The current tendency to reduce the dimensions of the mechanical transducer in the new generation of micro / nanoelectromechanical systems (M / NEMS) makes the reading system used to detect the movement of the structure (eg a lever) a topic of interest, remaining practically excluded optical systems due to the aforementioned problems.

An example of a sensor can be seen in Fig. 1 based on lever according to the state of the art, in which the mass sensor 10 comprises a lever 11 that resonates parallel to a static electrode 12 that allows excitation of the lever. Lever swing detection is achieved by detecting the capacitive current induced at the interface between the polarization electrode 12 and the lever 11, which is proportional to changes in the capacity formed by the electrode of polarization and lever.

This type of detection easily allows the monolithic integration of the M / NEMS system , avoiding the complexity and costs of similar systems using other techniques such as optics [J.Fritz et al. Translating biomolecular Recognition into Nanomechanics. Science, vol. 288, 316-318 (2000)] . Said electrostatic transduction, and due to the small dimensions of the mechanical elements, requires minimization of the parasitic capacity, since the magnitude of the current to be detected is proportional to the coupling capacity between the lever and the electrode. polarizer, which is usually in the range of 10 - 16 - 10 - 18 F.

The use of switched techniques for reading circuits is widespread for capacitive sensors such as accelerometers, since they are more insensitive techniques to parasitic capacities [N. Yazdi et al., "A generic interface chip for capacitive sensors in low-power multi-parameter Microsystems", Sensors and Actuators, vol. 84, pp. 351-361 (2000)] . Examples of use of such techniques are described in [BE Boser and RT Howe, "Surface Micromachined Accelerometers", IEEE Journal of Solid-State Circuits, vol. 31, no. 3 (1996)] and [C. Lu, M. Lemkin and BE Boser, "A Monolithic Surface Micromachined Accelerometer with Digital Output", IEEE Journal of Solid-State Circuits, vol. 30, no. 12 (1995)] . However, the use of these techniques in resonant systems where the frequency of the sensor signal is in the range of MHz (to increase the sensitivity of the sensor), is not directly extrapolated.

Thus, for example, in Fig. 2 you can see a reading system, according to the state of the art, for a sensor mass 10 based on lever 11 excited by a voltage of Vexc excitation applied to static sensor electrode 12 (see also Fig. 1), to cause the lever to resonate in parallel to the aforementioned electrode. Lever swing detection is achieved by reading the capacitive current induced in the interface between the electrode and the lever, said current being capacitive proportional to any change in the capacity formed by the electrode and the lever. Therefore, any change in the capacity will be reflected in the capacitive current.

The reading system of Fig. 2, intended to detect and amplify said possible changes in the capacity C (t) formed by the electrode 12 and the lever 11 resulting from the resonant movement of the lever, is based on the detection of the voltage in AC (alternating current) in the Vin node, resulting from the capacitive voltage divider formed by the capacity to be measured C (t) and the sum of all the capacities hanging from the Vin node, whose sum is called parasitic capacity and corresponds , according to Fig. 1, at Cd + Csb + Ci. Obviously, the reading system of Fig. 2 requires a correct polarization of the detection circuit 20 acting as a buffer , setting the DC voltage (direct current) of the Vin node. This polarization is achieved through the use of Rbias, which is an element with a resistance much higher than the impedance of the Cd capacity (associated with Rbias) at the frequency of interest, so that the signal does not experience any attenuation. In this way it is possible that the capacitive current circulates practically only by the capacity and does not circulate through the resistive element Rbias.

For all this, the sensitivity of the systems reading for known transducer elements is very limited

Description of the invention

Therefore, it is an objective of the present invention provide a reading system for an element transducer, with a very high sensitivity.

This objective is achieved in accordance with the claim 1 providing a reading system for a mechanical transducer element in which the polarization means are connected between the input node and the output node of amplification means, to polarize the input node connecting it to the output node.

In this way, a reading system is achieved of very high sensitivity when capacity is minimized strongly parasite because the capacity of the element becomes practically nil of polarization, that is, it is possible to reduce the contribution of the polarization element capacity on the input node of the voltage amplification means, by Miller effect, since the gain of the amplification means is approximately equal to unity.

Cd_ {Vin} = Cd (1 - G) \ approx 0

This configuration allows the integration of the displacement current using the intrinsic capacity of amplification media input, polarized through a feedback through the polarization element, being said highly resistive element.

On the other hand, the transducer element can be contained in a substrate, the substrate being connected to the polarization means through the output node.

Therefore, the voltage at the output node It is also used to strongly eliminate parasitic capacity between the mechanical structure (transducer element) and the substrate which contains said mechanical structure.

Csb_ {Vin} = Csb (1 - G) \ approx 0

Consequently, maximization of reading system sensitivity by reducing all parasitic capacities hanging from the input node. So, the system output voltage can be expressed approximately by the following equation:

Vo = Vexc \ frac {C (t)} {C (t) + Ci}

Preferably, the amplification means, such as a voltage amplifier, they can present a gain very close to the unit and a very high input impedance with a very low input capacity. In addition, it may be necessary that in its static transfer characteristic there is such a point that Vin (DC) is equal to Vo (DC).

The fact that the gain is unitary allows, by Miller effect, strongly reduce the contribution of the capacity between the substrate and the mechanical structure and the capacity of the polarization element on the input node, according to the configuration of the reading system of the invention.

According to a preferred embodiment of the invention, the polarization means may comprise an element highly resistive, of the order of 10 8 \ Omega or higher, such As a semiconductor device. Between the devices Known semiconductors mainly highlight diodes polarized at zero and transistors in subthreshold. The utilization of said highly resistive element is because the resistance of the polarization element must be much higher than the impedance of the capacity of said element at the frequency of interest, so that the signal does not experience any attenuation. The The use of a semiconductor device is because they have a high impedance without having to occupy a lot of area.

Preferably, the transducer element comprises a monolithic mechanical resonator, such as a lever.

Brief description of the drawings

For greater understanding of how much has been exposed Drawings are accompanied in which, schematically and only to non-limiting example title, a case study of realization.

In the drawings:

Fig. 1 shows a schematic diagram of a lever based mass sensor according to the state of the technique;

Fig. 2 shows a circuit-shaped diagram electrical of a reading system for a mechanical resonator of according to the state of the art;

Fig. 3 shows a circuit-shaped diagram electrical of a reading system for a mechanical resonator of according to the present invention;

Fig. 4 shows a circuit-shaped diagram electrical of a capacitive reading system for a lever that vibrates laterally excited electrostatically, according to the invention;

Fig. 5 shows a diagram in the form of an electric circuit of a unit gain voltage amplifier ( buffer ) with output stage (UGB_OS) comprised in the reading system of Fig. 3;

Fig. 6 and Fig. 7 show diagrams in the form of graphs of the measured transmission response (magnitude and phase) of the MEMS (Fig. 6) and the oscilloscope waveforms of the Vo and Vexc nodes for different frequencies (Fig. .7).

Description of preferred embodiments

The description of a preferred embodiment of the invention in which the system of reading is capacitive and is used to detect the oscillation of an integrated sub-micrometric scale lever monolithically according to a standard CMOS technology. The lever sub-micrometric constitutes the transducer element of an ultrasensitive mass sensor with resolutions in the range of We attograms them. The reading system presents a capacitance total parasite in the detection node as low as 11 fF. Bliss parasitic capacitance is used to integrate the current generated by the lever-electrode assembly (driver) which is excited electrostatically, resulting in a system High sensitivity reading with a bandwidth (BW) of almost 50 MHz (with a load capacity of 30pF).

In Fig. 3 the reading system of the described preferred embodiment is shown. The electrode 12 is excited by an excitation voltage Vexc, which causes the resonance of the lever 11 parallel to the electrode. The detection of the oscillation of the lever is achieved by detecting the capacitive current Ic induced at the interface between the electrode 12 and the lever 11, which is proportional to any change in the capacity C (t) formed by the electrode and the lever. To perform said detection, the reading system comprises a voltage amplifier 20 acting as a buffer , whose input node Vin (detection node) is connected to the output of the sensor 10, and a Rbias polarization element of the amplifier 20 , which is connected according to a feedback configuration, connecting the polarization element between the Vin input and the Vo output of the amplifier 20. Said polarization element is necessary to set the DC voltage (direct current) of the Vin node in order for the amplifier 20 It is correctly polarized. This polarization is carried out by any element that has a much higher resistance than the impedance of the Cp capacity (parasitic capacity of the reading system) at the frequency of interest so that the signal does not experience any attenuation. It is therefore necessary to integrate resistances of the order or greater than 1018 \ Omega. To obtain a high impedance without having to occupy a lot of area, semiconductor devices such as zero polarization diodes, sub-threshold transistors, etc. are normally used. The sensor 10 also comprises an N-well substrate containing the mechanical structure, that is, the lever-electrode assembly (driver).

The function of the reading system is to detect and amplify small variations in capacity (C (t)) resulting from the movement of lever 11 (mechanical resonator). Basically, the voltage detection is performed on the Vin node resulting from the capacitive voltage divider formed by the ability to measure C (t), formed by electrode 12 and the lever 11, and the sum of all the capabilities hanging from the node Vin (Cp), so Cp corresponds to Ci. This is because, with the polarization element feedback setting Rbias, the Cd capacity of said polarization element practically it is canceled by Miller effect, and the same happens for the Csb capacity formed between the mechanical structure and the substrate at connect the output node Vo to the polarization of the substrate that It contains the mechanical structure.

A further description will be made below. details, comparing the reading system of the invention (Fig. 3) and the reading system of Fig. 2 according to the state of the technique.

According to the scheme in Figure 2, the voltage of Vo output of buffer 20 can be expressed according to the equation next, referenced as 1, where Vexc is the voltage of excitation that is applied to electrode 12 in order to induce the structure at resonance, G is the gain of the amplifier 20, Csb is the parasitic capacity between the mechanical structure and the substrate, Ci represents the input capacity of the circuit and the connected with the mechanical structure and Cd is the ability of the polarization element.

(1) Vo = Vexc \ frac {C (t)} {C (t) + Csb + Ci + Cd} G

From the previous equation, the need to minimize parasitic capacity (Cp = Csb + Cd + Ci) if You want to increase the sensitivity of the detection system. Of the state from the technique it can be deduced that the value of Cd is sometimes higher than the value of the other capacities and therefore limits the sensitivity of this reading system.

In summary, the resistive element (Rbias) necessary to polarize the buffer limits the sensitivity of the reading system, since its associated parasitic capacity Cp is of the same order of magnitude as the intrinsic capacity of the buffer, so it increases directly or by Miller effect.

The electrical scheme is detailed in Figure 3 proposed to minimize parasitic capacity (Cp) and therefore Maximize the sensitivity of the detection system. This scheme It has the following characteristics:

to)

The voltage amplifier 20 is used, which performs the functions of buffer and has the following characteristics: gain very close to the unit, very high input impedance with a very low input capacity (Ci) and that in the transfer characteristic static there is a point such that Vin (DC) is equal to Vo (DC).

b)

He input node (Vin) is polarized by connecting it to the output of the amplifier (Vo) using the Rbias polarizing element, more specifically a semiconductor device (polarized diode in zero, sub-threshold transistors, etc.) offering resistance very high but at the same time has a parasitic capacity associated (CD). With this scheme it is possible to reduce the capacity contribution (Cd) over the input node (Vin) by Miller effect (equation 2), since the gain, G, is approximately equal to unity.

(2) Cd_ {Vin} = Cd (1 - G) \ approx 0

C)

The amplifier output voltage (Vo) is also used for 15 strongly eliminate parasitic capacity (Csb) by connection to the polarization of the substrate that contains the structure mechanics (equation 3).

(3) Csb_ {Vin} = Csb (1 - G) \ approx 0

Therefore, with the reading system of the Fig. 3 maximizes its sensitivity by reducing of all parasitic capacities hanging from the input node Vin In this case, the output voltage Vo can be expressed approximately by equation 4:

(4) Vo = Vexc \ frac {C (t)} {C (t) + Ci}

Thus, the reading system of Fig. 3 has the Next operation The lever 11 is excited by applying a voltage to electrode 12 (Vexc), which has an AC component (alternating current), to harmonically excite the lever, and a DC component (direct current) to prevent the lever from having have to move twice the frequency of the applied AC signal as well as to maximize the current generated by the movement of the lever. The capacitive current Ic generated can be expressed approximately as:

(5) I_ {C} (t) = \ frac {\ partial} {\ partial t} (C \ text {.} V) \ cong C_ {p} \ frac {\ partial V_ {AC}} {\ partial t} + V_ {DC} \ frac {\ partial C (t)} {\ partial t}

where C is the capacity formed by the electrode and the lever composed of a static component C_ {p} plus a variable component C (t) that reflects the capacity variations due to lever movement. He first term of equation (4) is a parasitic current generated by the AC voltage and the second term is the current of movement (I_ {M}) that reflects the oscillation of the lever.

Using an approximation uni-dimensional in which the displacement of the lever is the same in the entire length of the lever, it  obtains, for small displacements of the lever, the following equation, under harmonic excitation:

(6) I_ {M} = V_ {DC} \ frac {\ partial C} {\ partial t} = V_ {DC} \ frac {\ partial C} {\ partial z} \ frac {\ partial z} {\ partial t} \ cong V_ {DC} C_ {p} \ omega \ frac {A} {s_ {o}} cos (\ omega t)

where A is the amplitude of the deflection of the lever tip, C_ {p} and s_ {o} are the static capacitance and the distance between the lever and the electrode, V_ {DC} is the DC voltage drop in C_ {p} and \ omega is the frequency of oscillation.

If I_ {C} is integrated using C_ {i} according to circuit shown in Fig. 4, the output of the reading system V_ {o} can be approximated by:

(7) V \ cong \ frac {V_ {DC} A} {s_ {o}} \ frac {C_ {p}} {C_ {p} + C_ {i}} G

From the last equation it is possible deduce that a low value of the total capacitance in the node of detection (Ci) is mandatory to maximize the sensitivity of the reading system since the Vin voltage is proportional to 1 / Ci. A very low value of Ci = 11fF has been achieved by designing the system with the following characteristics:

-

The intrinsic capacitance on the Vin node (input capacitance of the circuit, capacitance of the lever-substrate assembly and connection to the circuit) is used as the integrated capacitor Ci instead of an explicit capacitor.

-

He voltage amplifier is polarized by a diode of DC polarization (D_ {Bias}) connected between the detection node Vin and node Vo1, as can be seen in Fig. 4. Since no there is DC current through the diode, this acts as an impedance of extremely high value (10 15 ohms). Said diode is designs in such a way that the parasitic capacity formed between the well and the substrate is not in parallel with C_ {i}.

-

He polarization diode, or other polarization element, can limit the sensitivity of the reading system since its parasitic capacitance, of the same order of magnitude as Ci, can increase Ci directly or by Miller effect. In that design, the parasitic capacity is reduced considerably using a feedback configuration that connects the diode between Vo1 and Vin nodes. Since the reading circuit is buffer 20 of unit gain (Vin / Vol \ approx 1), the increase of Ci by The Miller effect is very small.

Fig. 5 shows the reading based circuit in a level shifter topology independent of threshold values (M1 ... M5) with an output stage in configuration Follower by source (M8, M9). It is important to note that Transistor dimensions are in micrometers. Saying circuit has a low input capacitance (7.37fF) that is get using the input stage in follower configuration by source The circuit has two outputs: Vo1 which has a gain of 0.974 with respect to Vin and is used to polarize Vin through a diode and Vo that has the ability to charge high capacitances (30pF) with a relatively high BW that It allows the circuit test measurements.

The lever has a length of 10 µm, a width of 600 nm, a thickness of 925 nm and the separation between the lever and electrode is 750 nm. With these values, the value Theoretical C_ {p} is as low as 0.1 fF.

The system has been implemented in a technology CMOS standard. The mechanical structure is defined using the conventional CMOS process. In the end you get an encapsulated chip to which it is only necessary to connect the power and read its exit. To integrate it, as lever 11 has to present very small dimensions and you don't want to use tensions high, it is necessary, as described above, minimize all the parasitic electrical capacities that are generated by the integration system itself.

The structural layer is a metal layer (layer upper) and the sacrificial layer is the SiO_ {2} existing under the structure. After the standard CMOS process, a post-process to release the lever.

In short, the lever reading system Resonant submicrometer scale has been manufactured and integrated monolithically with the mechanical structure in a technology CMOS standard. Said reading system is based on a circuit of capacitive detection and presents a total parasitic capacitance at the detection node as low as 11fF, which results in a High sensitivity reading system, which is the best result currently obtained for a capacitive and comparable lever with the best results obtained by optical systems. All this demonstrates the possibility of the reading system according to the invention of detecting the oscillation of a scale lever sub-micrometric with a resonance frequency of around 6MHz.

Both Fig. 6 and Fig. 7 are presented the results corresponding to the response measures of Transmission (Vo / Vexc) of the 10 electromechanical system. The lever has been excited with an alternating voltage of 1.6 Vpp and a voltage of 18V continuous polarization.

These measures have been carried out in vacuum conditions (around 10-7 bars), under which The resonator has a Q factor of 351 and a frequency of 6,631 MHz resonance. The large phase shift near the resonance makes possible the self-excitation of the resonator and uses said system as an oscillator.

Although they have been described and represented specific embodiments of the present invention, it is evident that the subject matter expert may introduce variants and modifications, or replace the details with technically equivalent ones, without depart from the scope of protection defined by the attached claims.

Claims (8)

1. Reading system for a mechanical transducer element (11) comprising voltage amplification means (20) defining an input node (Vin) and an output node (Vo), at whose input node (Vin) connects the output of the transducer (11) and the amplification means (20) amplify the voltage at the input node (Vin) corresponding to the voltage at the output of the transducer element (11), and polarization means (Rbias) of said amplification means (20), characterized in that the polarization means (Rbias) are connected between the input node (Vin) and the output node (Vo) of the amplification means (20), to polarize the input node (Vin) connecting it to the output node (Vo). 2. System according to claim 1, characterized in that the amplification means (20) have a gain (G) very close to the unit. 3. System according to any one of claims 1 or 2, characterized in that the amplification means comprise a voltage amplifier (20). System according to any one of claims 1 to 3, characterized in that the polarization means comprise a highly resistive element (Rbias), of the order of 10 8 \ Omega or higher. 5. System according to claim 4, characterized in that the highly resistive element is a semiconductor device (RBias). 6. System according to any of claims 1 to 5, characterized in that the polarization means have a parasitic capacity (Cd) associated. 7. System according to any of claims 1 to 6, characterized in that the transducer element (11) is contained in a substrate (N-well), and the fact that the substrate is connected to the media (Rbias ) polarization through the output node (Vo). System according to any one of claims 1 to 7, characterized in that the transducer element comprises a monolithic mechanical resonator (11).