Microelectromechanical resonator

Abstract

To enable temperature stability and wear resistance, for example.SOLUTION: In a MEMS device having a substrate and a movable micromachined member, a mechanical structure secures the movable micromachined member to the substrate, thermally isolates the movable micromachined member from the substrate, and provides a conduction path to enable heating of the movable micromachined member to a temperature of at least 300°C.SELECTED DRAWING: Figure 1A

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to and incorporates by reference U.S. Provisional Patent Application No. 62 / 181,767, filed June 19, 2015, and U.S. Provisional Patent Application No. 62 / 183,689, filed June 23, 2015. This application incorporates by reference U.S. Provisional Patent Application No. 61 / 937,601, filed February 9, 2014, and U.S. Provisional Patent Application No. 14 / 617,753, filed February 9, 2015, respectively.

[0002] Technical Field TECHNICAL FIELD The disclosure herein relates to the field of microelectromechanical systems (MEMS), and more particularly to resonant MEMS structures.

[0003] BRIEF DESCRIPTION OF THE DRAWINGS Various embodiments disclosed herein are illustrated by way of example, and not by way of limitation, in the accompanying drawings in which: [Brief explanation of the drawings]

[0004] [Figure 1A] 1 illustrates various physical, electrical, and TCF industrial design aspects of an exemplary piezoelectric actuated MEMS resonator having at least two degenerately doped silicon layers. [Figure 1B] 1 illustrates various physical, electrical, and TCF industrial design aspects of an exemplary piezoelectric actuated MEMS resonator having at least two degenerately doped silicon layers. [Figure 1C] 1 illustrates various physical, electrical, and TCF industrial design aspects of an exemplary piezoelectric actuated MEMS resonator having at least two degenerately doped silicon layers. [Figure 2A] Examples of positive / negative first and second order frequency temperature coefficients (TCFs) are shown, as well as positive and negative zeroth order TCFs (ie, temperature independent frequency offsets). [Figure 2B]1 illustrates an embodiment of a resonant structure constructed from one or more materials that allows control of the primary and secondary resonator frequency temperature coefficients. [Figure 2C] 1 illustrates an embodiment of a resonant structure constructed from one or more materials that allows control of the primary and secondary resonator frequency temperature coefficients. [Figure 2D] 1 illustrates an embodiment of a resonant structure constructed from one or more materials that allows control of the primary and secondary resonator frequency temperature coefficients. [Figure 2E] 1 illustrates an embodiment of a resonant structure constructed from one or more materials that allows control of the primary and secondary resonator frequency temperature coefficients. [Figure 2F] 1 illustrates an embodiment of a resonator having degenerately doped semiconductor layers with non-uniform dopant concentration and / or type across the resonator body. [Figure 2G] 1 illustrates an embodiment of a resonator having a region within or on top of the resonator of locally deposited or patterned material having desired properties. [Figure 2H] 2B-2G and other embodiments disclosed herein. [Figure 2I] FIG. 2E shows a resonator embodiment in which the conductive layer is made thin enough to make its contribution to the TCF of the composite structure negligible. [Figure 2J] 1 illustrates an embodiment of a MEMS resonator having degenerately doped single crystal silicon layered with aluminum nitride sandwiched between top and bottom electrodes. [Figure 3] 1 illustrates an embodiment of a MEMS system in which a resonator is fabricated together with a temperature sensing element and a heater on a single substrate (the temperature sensing element and / or heater can optionally be eliminated in alternative embodiments). [Figure 4A] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4B]1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4C] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4D] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4E] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4F] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4G] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4H] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 4I] 1 shows an embodiment of a MEMS system in which a temperature-stable MEMS resonator is combined with additional active temperature compensation for improved frequency stability with temperature. [Figure 5A] 1 illustrates an embodiment of a degenerately doped silicon MEMS resonator integrated with one or more components of an active temperature compensation system. [Figure 5B] 5B illustrates an example of a temperature compensation operation in the embodiment of FIG. 5A. [Figure 6] 1 illustrates various options for integrating a temperature sensor in an oscillator system with dual silicon resonators. [Figure 7] 1 illustrates an exemplary process for fabricating a MEMS thermistor and a dual silicon MEMS resonator in adjacent regions of the same SOI substrate. [Figure 8] 1 illustrates an example interconnect configuration within a dual silicon resonator having an additional lightly doped (or undoped) single crystal silicon thermistor layer as part of the resonator structure. [Figure 9A] 1A-5B illustrate an embodiment of a degenerately doped semiconductor (DDS) resonator having one or more features described in connection with FIGS. 1A-5B. [Figure 9B] 9B shows an exemplary finite element model of the DDS resonator of FIG. 9A. [Figure 10A] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10B] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10C] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10D] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10E] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10F] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10G] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10H] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 10I] A non-exhaustive list of dual silicon resonator geometries and designs is presented. [Figure 11A]10A-10C illustrate embodiments of a dual silicon resonator having one or more metal layers or silicides to improve the conductivity of the electrodes and / or to act as a seed bed for improving the structuring of the material of the inner piezoelectric layer. [Figure 11B] 10A-10C illustrate embodiments of a dual silicon resonator having one or more metal layers or silicides to improve the conductivity of the electrodes and / or to act as a seed bed for improving the structuring of the material of the inner piezoelectric layer. [Figure 12A] 1B illustrates an alternative material stack having the three layers shown in FIG. 1A together with an additional metal layer. [Figure 12B] 1B illustrates an alternative material stack having the three layers shown in FIG. 1A together with an additional metal layer. [Figure 13A] 1 illustrates an embodiment of a symmetric dual silicon material stack that avoids surface tension imbalance and resulting structural warpage. [Figure 13B] 1 illustrates an embodiment of a symmetric dual silicon material stack that avoids surface tension imbalance and resulting structural warpage. [Figure 14] 1 illustrates exemplary micromachined members (or structures) within a microelectromechanical system (MEMS) that can be heated or superheated using various techniques and circuitry disclosed herein. [Figure 15A] 1 shows a profile view of an example MEMS device having a DDS resonator with structural interconnects that allow for Joule heating during fabrication. [Figure 15B] 10 illustrates a post-encapsulation joule heating MEMS embodiment in which conductive vias extend through a hermetic encapsulation lid configuration to make electrical contact with a conductive tethering structure. [Figure 15C] An alternative embodiment is shown in which conductive vias extend through the device substrate (eg, bulk semiconductor) to allow Joule heating current to flow through the DDS resonator. [Figure 16A] 1 illustrates an embodiment of a multi-functional tether structure, showing an example of a bulk tether. [Figure 16B]1 illustrates an embodiment of a multi-function tether structure, showing an example of a composite tether. [Figure 16C] 1 illustrates an embodiment of a multi-function tether structure showing an example of an extended path tether. [Figure 16D] 10 shows an example thermal profile achieved by conducting DC or AC Joule heating current through a movable micromachined member via an anchor and folded tether. [Figure 17] FIG. 10 illustrates another Joule-heated MEMS embodiment in which individual pairs of "flexible" tethers are provided to conduct Joule-heating current to a movable micromachined member, each tether pair thus forming a dedicated heating port. [Figure 18A] An exemplary capacitively coupled heating configuration is shown in which electrodes functioning as drive and / or sense electrodes serve dual purposes as radio frequency (RF) energy inputs that drive AC Joule heating current within a movable micromachined member. [Figure 18B] An exemplary capacitively coupled heating configuration is shown in which electrodes functioning as drive and / or sense electrodes serve dual purposes as radio frequency (RF) energy inputs that drive AC Joule heating current within a movable micromachined member. [Figure 19A] 1 illustrates a Joule heating embodiment in which energy is supplied conductively or capacitively to individual layers of a multi-layer MEMS resonator. [Figure 19B] 1 shows an alternative heating configuration in which Joule heating is primarily provided within the patterned electrode layer of a dual silicon resonator. [Figure 20] 19B illustrates another embodiment in which an RF energy source is coupled across the outer layer of the movable micro-machined member shown in FIG. 19A. [Figure 21] 1 illustrates an encapsulated or chip-scale MEMS package having a MEMS die (eg, having a movable micromachined member formed thereon) and a control die. [Figure 22]23 illustrates one embodiment of a programmable heating controller that can be used to implement the programmable heating control circuit of FIG. 22. [Figure 23] 1 illustrates a configuration in which multiple heating power drivers within an external heating controller can be coupled to individual singulated or on-wafer MEMS devices. [Figure 24A] 22 and 23 show example heating power profiles that may be generated within an external or in-situ heating controller or any other possible heating source, along with example temperatures generated within a movable micro-machined member. [Figure 24B] 1 illustrates exemplary power pulse shaping that may be applied to achieve controlled cooling and / or heating of a MEMS structure. [Figure 25] 1 illustrates an exemplary process for a packaged MEMS device in which an oven reflow operation is performed to merge two separate terminals that form a heating port into a single operable control terminal. [Figure 26] 1 illustrates an example heating port configuration in a packaged MEMS device having two front terminals and one back terminal to allow for joule heating / superheating. [Figure 27A] 27 illustrates an exemplary circuit model corresponding to the packaged MEMS device of FIG. 26 during heating / superheating operation. [Figure 27B] 27 illustrates an example circuit model corresponding to the packaged MEMS device of FIG. 26 in run-time operation. [Figure 28A] An exemplary electrical interconnection between an encapsulation level through silicon via (TSV) and a piezoelectric actuated resonator electrode is shown, which allows both package level terminal interconnection of the MEMS resonator (i.e., in a subsequent packaging step) and post-encapsulation Joule heating. [Figure 28B] An exemplary electrical interconnection between an encapsulation level through silicon via (TSV) and a piezoelectric actuated resonator electrode is shown, which allows both package level terminal interconnection of the MEMS resonator (i.e., in a subsequent packaging step) and post-encapsulation Joule heating. [Figure 29A] 1 illustrates an alternative terminal reduction scheme in which three or more encapsulation level terminals are merged into an enclosure within the package housing to expose only two electrically independent package level terminals. [Figure 29B] 1 illustrates an alternative terminal reduction scheme in which three or more encapsulation level terminals are merged into an enclosure within the package housing to expose only two electrically independent package level terminals. [Figure 30] 1 shows heating of a MEMS resonator via a laser beam that is preferentially absorbed by the DDS in relation to a lid and substrate that are part of a hermetic encapsulation. [Figure 31A] 1 shows the variation of electroactive dopant concentration within the DDS as a function of temperature during heating cycles. [Figure 31B] 1 shows the cooling rate dependence of the electrically active dopant concentration. [Figure 32A] 1 shows a reduction in room temperature resonant frequency with increasing electroactive dopant concentration. [Figure 32B] 1 shows the dependence of the room temperature resonant frequency of a MEMS resonator containing a DDS on both time and temperature of a heating cycle in which the resonator is then rapidly cooled. [Figure 33] 1 shows a sequence of heating pulses applied to a MEMS resonator, the corresponding average temperature of the MEMS structure in each cycle, and the resulting resonant frequency of one of the MEMS resonator eigenmodes at or near room temperature after each heating pulse. [Figure 34] An algorithm is presented for trimming the frequency of one or more resonant modes of a MEMS structure at one or more temperatures in an automated, closed-loop manner. [Figure 35] 10 shows exemplary frequency trimming results in which the thermal power and / or time is increased over a series of repeated heating pulses until the room temperature frequency falls within an acceptable tolerance around the target frequency. [Figure 36]Utilizing the reversible property of temperature-activated frequency tuning, an example frequency trimming result is shown in which the heating power is reduced after three iterations to increase the resonant frequency after initially overshooting the frequency trim target. [Figure 37A] Shown are two vibration modes of interest, mode 1 and mode 2, in this particular example a MEMS structure, a disk ring gyroscope. [Figure 37B] An array of electrodes is shown surrounding the MEMS structure of FIG. 37A, which may also be utilized for capacitively coupled Joule heating, in addition to potential uses in normal device operation. [Figure 37C] 37B shows radio frequency (RF) input applied to a subset of electrodes to cause non-uniform heating of the MEMS structure in order to preferentially tune the frequency of Mode 1 of the MEMS structure of FIG. 37A. [Figure 37D] 37B shows RF input applied to different subsets of electrodes to preferentially adjust the frequency of Mode 2 of the MEMS structure of FIG. 37A to obtain a desired frequency relationship between the two modes. [Figure 38] We demonstrate tuning the frequency relationship between two modes of a similar MEMS structure based on activation of a subset of heating terminals, except that Joule heating is achieved through terminals directly coupled to the structure, thereby allowing for the possibility of DC heating current. DETAILED DESCRIPTION OF THE INVENTION

[0005] Detailed Description Various embodiments herein disclose temperature-stable, wear-resistant resonators formed by a material stack having one or more degenerately doped silicon layers and piezoelectric material layers. In some implementations, the piezoelectric material layer ("piezoelectric layer") is sandwiched between a degenerately doped single-crystal silicon "core" layer and a degenerately doped polycrystalline layer, where the outer silicon layers act as electrodes for conducting drive / sense signals to and from the piezoelectric layer, thereby obviating the traditional metal electrode layers and their undesirable aging characteristics (e.g., wear-hardening over time). Furthermore, as described in more detail below, the layer thickness ratios, crystal orientations, mode shaping, and / or degenerate dopant concentrations / types of (at least) the single-crystalline layers may be engineered to substantially zero or null at least the first-order (linear) and second-order (parabolic) temperature coefficients of frequency (TCF) in the material stack as a whole, where the inherent parabolic TCF of the piezoelectric layer is substantially nulled by an oppositely polarized parabolic TCF engineered in, for example, the degenerately doped single-crystalline silicon layer. Thus, in addition to serving a dual purpose as both a sense / drive electrode and a resonator bulk layer, the degenerately doped single-crystalline silicon layer provides engineering “knobs” for resulting in a resonant MEMS structure with a temperature-stable resonant frequency (e.g., a zero or near-zero combined TCF) over a desired operating temperature range (e.g., −40° C. to +85° C.). Similarly, the degenerately doped polysilicon layer functions as a wear-resistant electrode (and can therefore be considered part of the sense / drive component along with the piezoelectric layer) and can also enable multiple degrees of TCF engineering.In still other embodiments, one or more in-situ temperature sensing elements are provided (e.g., implemented within the resonator package, within the encapsulation chamber, or possibly within the resonator structure itself) to enable temperature-regulated frequency pulling and / or post-resonator output frequency adjustment to flatten the net resonator TCF (i.e., the sum of the 1st, 2nd, ..., nth order TCFs) and / or to compensate for aging effects, or possibly to allow run-time frequency adjustment (e.g., by enabling closed-loop heating to a desired operating temperature which affects the 0th order TCF). These and other features and embodiments are described in further detail below.

[0006] 1A-1C illustrate various physical, electrical, and TCF engineering design aspects of an exemplary piezoelectric actuation MEMS resonator 100 having at least two degenerately doped silicon layers, referred to herein as a "dual silicon" resonator. Referring first to FIG. 1A, the physical stack of materials shown in cross section A-A' includes a degenerately doped (DD) single-crystal silicon (SC-Si) core, a piezoelectric layer (dielectric), and a degenerately doped polycrystalline silicon (DD poly) layer. The dopant concentration in the outer silicon layers is selected (e.g., 1E18 atoms / cm) to provide low-loss conductivity and thereby enable the silicon layers to function as electrodes for establishing a time-varying piezoelectric actuation voltage across the piezoelectric layers, thereby avoiding conventional metal layers that tend to harden over time and shift the resonator frequency in an undesirable manner. 3 , 1E19 atoms / cm3, 1E20 atoms / cm 3 , 2E20 atoms / cm 3In the particular embodiment shown, top and bottom electrodes ("top" and "bottom" refer arbitrarily to the orientation of the drawing, as the structure may be flipped or otherwise reoriented during deployment, use, etc.) implemented by poly and single crystal silicon layers, respectively, are coupled to a piezoelectric actuation "drive" signal (V sustain ) and a piezoelectric output signal (V sense ) to external package contacts through a mechanical tether and tethering structure to enable external sensing of the resonator. As shown, the resonator material stack can be electrically modeled by a plate capacitance (piezoelectric layer) coupled to the drive and sense nodes through low-loss conductive paths.

[0007] 1B shows an example cross section through the anchor, tether, and body of a dual silicon resonator in more detail. As shown, the piezoelectric layer is mounted by a thin aluminum nitride (AlN) film extending through the tether and anchor, along with two degenerately doped silicon layers. Although shown as decoupled elements, the tether, anchor, and resonator body are physically coupled to one another, with the tether functioning as a spring-like structure to allow mechanical movement of the resonator ("movable micromachined member") that would otherwise be released from and suspended relative to the surrounding field area and underlying substrate or insulator layer (alternatively, or in addition, the anchor and one or more tethering structures may be disposed below the resonator body). Thus, a conductive path can be formed through the anchors and tethers to generate an electrostatic potential across the aluminum nitride piezoelectric layer, and thereby actuate the resonator, thereby generating oscillating mechanical motion thereof in one or more mechanical resonant modes (e.g., elongation, breathing, Lamellar, flexure, bulk-acoustic, or any other possible MEMS resonant mode—in-plane or out-of-plane, etc.). For example, and without limitation, aluminum nitride piezoelectric layers can be fabricated using zinc oxide (ZnO), lead zirconate titanate (Pb[Zrx Ti 1-x ]O3, 0≦x≦1), lithium niobate (LiNbO3), gallium nitride (GaN), indium nitride (InN), scandium aluminum nitride (ScAlN), quartz (SiO4), and the like may be used to implement the piezoelectric layer in the embodiment of FIG. 1B and all other embodiments disclosed herein.

[0008] In the context of TCF engineering, degenerately doped single-crystal silicon incorporates at least two authorities / degrees of control within a single layer: crystal orientation and dopant concentration. In the context of depositing a piezoelectric thin film on heavily doped single-crystal silicon, additional authorities are available. Specifically, the thickness ratio ("x / y") of the degenerately doped single-crystal silicon layer and the aluminum nitride piezoelectric layer can be selected to yield substantially matched, but oppositely signed, second-order TCFs within these two layers, thereby nullifying or substantially attenuating their combined parabolic contribution to the net resonator TCF, and also providing some control (authority) over the third-order resonant TCF. Preferred crystal orientation within the polysilicon thin film can also influence the first- and second-order TCFs. Thus, as shown in FIG. 1C, by varying the crystal orientation, dopant type and layer thickness of the degenerately doped single-crystalline silicon layer at a particular doping concentration, the thickness of the aluminum nitride layer, and / or the thickness of the degenerately doped polysilicon layer at a particular dopant concentration, at least the first and second order TCFs of the dual silicon piezoelectric actuated resonator of FIGS. 1A and 1B can be engineered (controlled, manipulated). In some embodiments, for example, various TCF levels are engineered to provide a substantially zero net TCF (e.g., frequencies within 50 ppm, 20 ppm, 10 ppm, or less of a target) over a desired operating range (e.g., −40° C. to 85° C.). More generally, such multi-order compensated resonators (i.e., nulling not only the first order TCF but also second order and possibly third and higher order TCF terms) can: Nulling or otherwise attenuating (e.g., to zero, substantially zero, or other negligible level) the first-, second-, third-, and fourth-order TCF terms, or any two of them; exhibiting a non-zero value of linear TCF specifically selected to partially or fully compensate (or nullify or cancel) third-order TCF and thereby reduce absolute frequency variation over a particular temperature range; exhibit one or more local extrema in frequency that are temperature dependent (i.e., have temperatures at which local minimum or maximum frequency occurs), also referred to as "turnover temperatures"; - Exhibits turnover at nominal operating temperature, The design may be engineered as follows.

[0009] Also, as a departure from TCF compensation schemes that cumulatively apply material layers to compensate for individual TCF terms (such as one or more material layers to compensate for first-order TCF and one or more other material layers to compensate for second-order TCF), embodiments of temperature-engineered resonant structures disclosed herein include semiconductor layers or regions that are engineered to compensate for both the linear TCF and one or more higher-order TCFs, thereby enabling the construction of "short stack" MEMS resonators composed of a reduced number of material layers relative to the layer / TCF implementation. In one example, the linear TCF of a resonator fabricated from an anisotropically degenerately doped semiconductor (DDS), such as single-crystal silicon, is adjusted by rotating the orientation of the resonator shape relative to the underlying crystal axes. In a second example utilizing anisotropically degenerately doped semiconductors, the linear TCF of a resonator is tuned by changing the resonator shape and / or mode shape. As another example, in some embodiments, a semiconductor material is doped sufficiently heavily so that the polarity of the second-order TCF (i.e., the sign of the second-order TCF) of a resonator constructed from that material is reversed relative to the second-order TCF of a resonator constructed from a more lightly doped version of that semiconductor (e.g., resulting in a positive second-order TCF, as opposed to the negative second-order TCF exhibited at a relatively low dopant concentration). The opposite polarity second-order TCF may be exploited to partially or entirely nullify the second-order TCF of one or more other materials in a composite resonator embodiment (e.g., a piezoelectric material having a second-order TCF polarity opposite to that of the degenerately doped semiconductor material) over a given temperature range, thereby producing a resonator with a reduced magnitude of the second-order TCF. As described, resonant structures with engineered linear and higher-order TCFs can be fabricated solely from degenerately doped semiconductors, or degenerately doped semiconductors may be deployed within composite resonant structures to enable engineered temperature stability.For example, composite structures suitable for piezoelectric micromechanical resonators are disclosed below, which address the issues of frequency stability with temperature and the engineering design of linear and higher order TCFs. As described, such composite structures offer high quality factors, low hysteresis with temperature, low frequency aging with time, compatibility with batch microfabrication methods, small form factors, low sensitivity to shock and vibration, and other desirable features.

[0010] The resonant structures described herein can be fabricated, in whole or in part, from degenerately doped single-crystal or polycrystalline semiconductors (e.g., silicon, germanium, diamond, carbon, silicon carbide, compound semiconductors such as silicon germanium, etc.), or by using combinations of such semiconductors. Pure and lightly doped semiconductors are insulating materials at low temperatures. When semiconductors are doped with certain impurity atoms above some dopant concentration, the semiconductor will exhibit metallic or highly conductive behavior even at low temperatures (e.g., as in single-crystal silicon near 0 Kelvin). Such semiconductors are said to be "degenerately doped." For example, in single-crystal silicon, this is 4E18 atoms / cm 3 This can occur at phosphorus doping levels above 100°C. More generally, the onset of degeneracy varies with semiconductor and dopant. In addition to changes in conductivity, various material properties of semiconductors vary with doping level, as well as the temperature dependence of various material properties. In particular, resonators fabricated with degenerately doped semiconductors can exhibit significantly different TCFs than similar resonators constructed with relatively lightly doped versions of the semiconductor material. In some cases, it is even possible to reverse the sign (or polarity) of one or more TCFs by changing the doping level of the semiconductor used as a structural material in the resonator. These temperature-dependent changes are exploited in some resonator embodiments to enable advanced engineering of targeted temperature coefficients.

[0011] In certain embodiments of the resonator composite structures disclosed herein, two of the constituent materials are a degenerately doped semiconductor (DDS) and a piezoelectric material. This composite structure, which is an example of a DDS resonator, can be engineered to simultaneously achieve target values or ranges of values for two or more of the TCFs of a particular resonant mode. In one embodiment, for example, the composite structure is constructed from degenerately doped silicon (a semiconductor) and aluminum nitride (a piezoelectric material—as noted above, others may be used) and is engineered so that the first- and second-order TCFs (i.e., linear and quadratic TCFs, respectively) of a particular resonant mode of the overall structure are both within a specified tolerance of zero, thereby resulting in a temperature-insensitive or temperature-compensated resonator.

[0012] In addition to the two main constituent materials, additional materials may be present within the composite structure. Also, in particular, other conductive materials (which may also be degenerately doped), such as metal or another semiconductor layer, may be included to function as an additional electrode (as described below, the DDS layer can function as another electrode within the composite structure). Examples of suitable electrode materials include, but are not limited to, heavily doped silicon, silicide, tungsten, molybdenum, titanium, or aluminum. The term electrode is used herein, without limitation, to mean a conductive material used to establish an electric field for the conversion of electrical energy to mechanical energy or mechanical energy to electrical energy. It should be noted that a layer applied as an electrode may also serve other functions, such as, for example and without limitation, piezoelectric or heating functions.

[0013] The embodiments disclosed herein address many or all of the above-mentioned problems and challenges in resonator performance through the engineering of multiple parameters or design degrees of freedom of structure with reduced sensitivity to temperature while providing piezoelectric coupling for some structures, high quality factor, low hysteresis, low aging, compatibility with batch micro-fabrication, small form factor, insensitivity to shock and vibration, etc.

[0014] In the case of mechanical resonators, the natural frequency is primarily determined by the mass density and stiffness of the material from which the resonator is constructed. Changes in material stiffness due to changes in ambient temperature are a major concern in resonator design and manufacture, as this will change the natural frequency of the resonator. Changes in material shape due to thermal expansion or contraction are also a concern, as this will also change the natural frequency of the resonator.

[0015] For educational purposes, this can illustrate the principles of TCF engineering design to examine a simple model. One such model is when the resonator structure is constructed from thin layers and only material motion on a single axis is considered. In this simple example, the stiffness of the material is described by a single numerical value, the effective modulus. This is a simplification of the physical case where all dimensions of motion need to be considered and the stiffness of the material can be described by a tensor. In this simple model, the modulus of an anisotropic material depends on the orientation of the material.

[0016] In the case of composite structures, a simple model for the nth order frequency temperature coefficient is the weighted average of the contributions of all the components of the resonator, which in the case of a stack of thin films (or layers) of materials can be written as:

number

number

number

number

[0017] Equation (2) shows that the first requirement for engineering the temperature coefficient of frequency of a composite resonator is to use materials that have a desired ensemble of values. For example, λ n If it is desired that is zero, then at least one

number

number

[0018] Moreover, equation (2) is

number

number

[0019] Despite the traditional emphasis on linear TCF control, analysis shows that composite resonator performance can be significantly enhanced through control over at least the first two temperature coefficients. As discussed above, resonator implementations with control over at least the first two frequency temperature coefficients can be constructed from aluminum nitride and degenerately doped single crystal silicon. Such resonators can be compatible with piezoelectric transduction and can have other advantageous properties, including, without limitation, a high quality factor (Q), low hysteresis with temperature, low frequency aging with time, compatibility with batch microfabrication, a small form factor, and insensitivity to shock and vibration.

[0020] In addition to controlling at least first and second order TCF, the following non-exhaustive list of criteria is applied in designing selected resonator embodiments disclosed herein. The target frequency temperature coefficient of the overall structure is achieved by combining the temperature coefficients of the individual materials, e.g., if the target first-order TCF is zero or near zero and at least one component material has a substantially positive first-order TCF, then at least one other component material is engineered and / or selected to have a negative first-order TCF. There are N design parameters to allow control over N temperature coefficients. The combination of design parameters has sufficient design authority to allow for a solution within the design space defined by the manufacturing and design constraints.

[0021] Aluminum nitride resonators typically have negative linear and quadratic TCFs. The temperature coefficient of thin-film polycrystalline aluminum nitride is less dependent on the thin-film structure. Similarly, resonators constructed from non-degenerate single-crystal silicon have negative linear and quadratic TCFs, and the TCF tends to be less dependent on crystal orientation.

[0022] The linear TCF of a resonator constructed from degenerately doped single-crystal silicon can be positive or negative, depending on the crystal orientation, doping level, and mode shape. Thus, crystal orientation constitutes a design parameter (or degree of design freedom) that can be adjusted to control the linear TCF term. The second-order TCF of a resonator constructed from degenerately doped single-crystal silicon can be positive or negative, depending on the dopant concentration, crystal orientation, doping level, and mode shape. Figure 2A shows examples of such positive / negative first- and second-order TCFs, as well as positive and negative zeroth-order TCFs (i.e., temperature-independent frequency offsets).

[0023] For example, through manipulation of orthogonal design parameters such as crystal orientation and dopant concentration, it becomes possible to independently tune (i.e., control and potentially null) both the first- and second-order (linear and quadratic) TCF characteristics of a degenerately doped single-crystalline silicon layer. Also, while polycrystalline silicon resonators can be degenerately doped to achieve a range of linear and quadratic values, one degree of design freedom can be lost if the material lacks a primary crystal orientation.

[0024] 2B-2E show an exemplary embodiment of a DDS resonator along with a plan view of the resonator shown in FIG. 1a, along with optional electrode arrangements within and adjacent to the resonator structure.

[0025] Referring first to FIG. 2B, a resonant structure including at least one degenerately doped semiconductor layer ("DDS resonator") is disposed between two electrode structures (e.g., used to drive and sense the resonator) and with one or more anchor points. In the illustrated embodiment, spring-like members or "tethers" extend from opposite sides of the resonator body to anchors within the surrounding material field to establish dual anchor points, although alternative embodiments can implement more or fewer anchor points. Also, while a rectangular or oblong resonator shape is illustrated, DDS resonators can be fabricated in any feasible shape and / or aspect ratio.

[0026] The electrodes on either side of the DDS resonator are shown in dashed outline to emphasize their optional nature (a convention that also applies to other figures herein; the absence of dashed lines should not be interpreted as meaning that a given element or structure is required). One or more electrically isolated electrode regions may also be implemented within a given resonator layer, as indicated by the outlined T-shaped region within the resonator body. In alternative embodiments, one or more alternatingly shaped electrode regions (i.e., having a different shape than shown) may be utilized, and the resonator body itself may also be utilized as an electrode.

[0027] In the case of a single-layer degenerately doped silicon resonator, resonator motion can be sensed electrostatically or piezoelectrically with appropriate electrical connections. Figures 2C-2E show cross sections (i.e., across the resonant structure at line A-A' in Figure 2B) of example material stacks, including a monolithic (single-layer) stack in Figure 2C, a two-layer material stack in Figure 2D, and a three-layer material stack in Figure 2E. As discussed below, in alternative embodiments, additional material stacks and / or stacks of materials different from those shown in Figures 2C-2E may be present. As discussed above, because a single layer of uniform degenerately doped silicon has at least two design parameters (crystal orientation and dopant concentration), a single-layer resonator (Figure 2C) constructed from uniform degenerately doped single-crystalline silicon can be engineered to have zero or near-zero first- and second-order TCFs through a combination of manufacturing process and design. Furthermore, the dopant concentration need not be uniform. This allows for an arbitrary number of design parameters (i.e., "knobs" or degrees of freedom for manipulating resonator performance through design). It may also be advantageous to create one or more regions within or on the resonator with different dopant concentrations and / or different dopants, as shown, for example, by the different degrees of shading in FIG. 2F. Regions within or on the resonator could also be created with locally deposited or patterned layers of material with desired properties, as shown in FIG. 2G. For example, it may be advantageous to place these regions within areas of high stress within the resonator's vibration mode shape, thereby allowing control over first-, second-, third-, or even higher-order TCF values.

[0028] Resonators can also be engineered to have non-zero, but controllable, first- and second-order TCFs to achieve specific design intent. For example, the first-order TCF can be designed to minimize the total frequency variation with temperature by compensating for third-order and other odd-numbered higher-order TCFs. As another example, the first-order TCF can be designed to adjust the temperature at which the change in frequency with temperature reaches an inflection point, local minimum, or local maximum. The temperature at which a resonator reaches a local minimum or maximum frequency is commonly referred to as the turnover temperature. The resonator TCF may also be designed to disable the TCF associated with its sustaining circuit (i.e., the circuit that maintains the mechanical motion of the resonator) or oscillator system. As a final example, the first- and second-order coefficients can be selected to be relatively insensitive to angle and dopant concentration for improved manufacturability. Figure 2H shows an example of such an engineered TCF.

[0029] Referring again to FIG. 2D, a two-layer resonator can be constructed from degenerately doped silicon and an additional thin film. If the silicon layer is single crystalline, this structure has at least three design parameters: the crystal orientation and dopant concentration described above, plus the ratio of silicon thickness to the thickness of the additional thin film. Thus, a two-layer resonator from degenerately doped silicon and an additional thin film can have the ability to control three TCFs. In some embodiments, including that shown in FIG. 1C, the additional thin film is a piezoelectric material (e.g., aluminum nitride), but the additional thin film (or other material layer) can instead be any semiconductor, insulator, or metallic material selected for its TCF coefficient or another desirable mechanical or electrical property.

[0030] Additionally, a three-layer resonator can be formed as shown in FIG. 2E (i.e., with or without the electrically isolated electrode regions described). In one embodiment, such a three-layer resonator includes a degenerately doped single-crystal silicon layer coated in aluminum nitride (to form the piezoelectric layer) and an additional conductive layer (e.g., a metal layer or another degenerately doped polysilicon electrode layer). As described above with reference to FIG. 2C, the primary and secondary TCFs can be controlled by varying the crystal orientation of the underlying (single-crystal) silicon at a particular doping level and thickness, the thickness of the aluminum nitride, and the thickness of the polysilicon at a particular dopant concentration. A preferred crystal orientation within the polysilicon thin film can also affect the primary and secondary TCFs. Alternative sets of design parameters can be selected to engineer the TCF of a thin-film stack, and the previous example is only one of many possibilities. In addition, as shown in Figure 21, one of the layers in a three-layer resonator can be made thin enough so that its contribution to the TCF of the composite structure is minimized (e.g., negligible or otherwise attenuated) and the compensation problem is reduced to that of the two-layer resonator case. For example, the top layer can be implemented by a thin conductive metal or semiconductor layer instead of a degenerately doped polysilicon electrode layer.

[0031] Additional layers can be added to the stack. One example, shown in Figure 2J, includes degenerately doped single crystal silicon with aluminum nitride sandwiched between a top electrode and a bottom electrode. The electrodes can be fabricated from any conductive metal or semiconductor thin film, such as polysilicon, aluminum, molybdenum, titanium, tungsten, or a silicide formed using a metal and silicon.

[0032] In various embodiments, a single MEMS system 200 can include multiple elements fabricated together on the same substrate as a DDS resonator. Referring to FIG. 3 , for example, MEMS system 200 can include a DDS resonating element 201 (i.e., one form of MEMS resonator), one or more temperature-sensing elements 203 (“temperature sensors”), and one or more heating elements 205 (“heaters”). The temperature-sensing elements may be used as part of an active temperature compensation system. One example of a temperature-sensing element is a thermistor, which has a temperature-dependent electrical resistance. Heaters may be included for initial calibration of the resonator frequency stability or to maintain the MEMS system at a substantially constant temperature despite fluctuations in the ambient temperature. Optionally, the temperature sensors and / or heaters may be eliminated from MEMS system 200. As discussed above (and in more detail below), the resonator structure may be used as a sensor instead of a frequency reference. Examples of alternative resonator applications include filters, gyroscopes, accelerometers, pressure sensors, microphones, magnetometers, and mass sensors.

[0033] The DDS resonators described herein may be deployed with or without complementary temperature compensation, thereby achieving a purely passive temperature compensation scheme or, for example, a combination of passive and active temperature compensation. While active electrical compensation circuits, by definition, increase the power consumption of a system, a combination of passive and active compensation (e.g., a DDS resonator in combination with an active compensation circuit) may enable the achievement of stability targets with less power than active compensation alone, or greater stability than could be achieved by either scheme alone. Passive mechanical temperature compensation is possible through material selection and structural design in both monolithic and composite resonator structures.

[0034] The DDS resonators disclosed herein may be combined with additional elements to form systems with improved temperature stability and / or other useful functions. In FIG. 4A , for example, DDS resonator 301 is combined with sustaining circuit 303 to form an oscillator. The resonant frequency of the oscillator may be altered in a temperature-dependent manner (i.e., as shown in FIG. 4A , a temperature signal from sensor 305 is received within frequency altering element 307, which then provides a temperature-dependent resonant frequency control signal to DDS resonator 301 and / or sustaining circuit 303) to provide a resonant frequency with improved temperature stability compared to that achievable with the DDS resonator alone. For example, a temperature-dependent electrostatic field may be applied directly to the DDS resonator using a control electrode fabricated with the resonator. FIG. 4B shows an example of this scheme, where the electrodes are formed by the degenerately doped silicon (DDSi) layer (electrode 2) and conductive layer (electrode 1) of the three-layer resonator embodiment described above (i.e., DDSi, AlN, and conductive layer, although other resonator structures / materials may be used in alternative embodiments), and a temperature-dependent electrostatic field is formed by a time-varying and temperature-dependent difference Ve1-Ve2 between the potentials of the two electrodes.

[0035] As another example, if one component of the resonator has a large piezoelectric response, the mechanical stress on the resonator can be modified in a temperature-dependent manner to adjust the resonator frequency. Figure 4C again shows an example of such a configuration in the context of the three-layer DDSi / AlN / conductive material resonator described above, although other resonator structures and / or materials may be used.

[0036] For example, in another embodiment shown in FIG. 4D, a capacitive element may be included in the resonator sustaining circuit 321, and its capacitance may be varied in a temperature-dependent manner, thereby providing a variable capacitance element 323 that may be used, for example, to tune the frequency of an oscillator system (i.e., a system including the DDS resonator 301 and sustaining circuit 321).

[0037] In an alternative embodiment shown in Figure 4E, the resonant frequency of the DDS resonator 301 is output from an oscillator (formed at least in part by the DDS resonator 301 and sustaining circuit 303, as shown) and then modified by a frequency-modifying element 331 within the resonator system. For example, in a more specific implementation shown in Figure 4F, a temperature-dependent signal (e.g., from a temperature sensor shown at 305 in Figure 4E) can be provided to a fractional-N phase-locked loop (an example of a frequency-modifying element) to produce an output frequency that is much more stable with temperature than the resonator alone.

[0038] As shown in FIG. 4G, any of the oscillator systems shown in FIGS. 4A-4F (or other oscillator systems utilizing DDS resonators) can include a programmable / storage circuit 350 within which TCF coefficients, control settings, or other information can be stored. The TCF coefficients recorded or stored within such oscillator systems can be based on the characteristics of individual resonators (301) or groups of resonators. In addition, the coefficients can be based on the characteristics of individual sustain circuits (303) or groups of sustain circuits. For example, the TCF behavior of an oscillator or resonator can be determined by recording frequency as the temperature is swept, as shown at 355, or by measuring room temperature properties (e.g., resistivity, frequency, etc.) that are predictive of time-dependent behavior.

[0039] The programmable / storage circuitry that may store the TCF coefficients or other parameters informing temperature-dependent behavior may include any operable on-chip or off-chip memory element, such as, for example and without limitation, a register or other volatile or non-volatile memory element, including, without limitation, a one-time-programmable (OTP) memory, an electrically programmable read-only memory (EPROM), a flash memory, etc.

[0040] The TCF coefficient, or other parameter indicative of temperature-dependent behavior, may also be stored as a resistance value. As shown in FIG. 4H, for example, a MEMS resonator die may include at least one resistor whose resistance value may be trimmed by a laser (or other method, such as thermal melting) to record a value related to the device's behavior with temperature. Thermistors and / or heaters may also be used to store information in this manner. For example, by using a thermistor or heater for this purpose, additional bond pads or electrical interconnects may be avoided.

[0041] The stored temperature behavior information may be used by the oscillator system to improve its frequency stability. The system can read the stored information, combine it with temperature measurements, and apply corrections to the resonator or oscillator system frequency. This operation is shown, for example, in the embodiment of FIG. 4I, where frequency modification element 375 receives a temperature signal from sensor 305 and reads temperature-based correction data (“data”) from programmable storage 350. A frequency correction signal corresponding to the correction data is then output to the resonator, a maintenance circuit, and / or an output conditioning circuit (e.g., the PLL described above with reference to FIG. 4F).

[0042] FIG. 5A shows one embodiment of a MEMS configuration having a DDS resonator 401 integrated with one or more components of an active temperature compensation system. The DDS resonator may be implemented according to any of the resonator embodiments described herein and includes one or more layers (or other depositions or configurations) of degenerately doped semiconductor material. In the illustrated embodiment, the temperature compensation circuitry includes one or more temperature sensing elements 403 (e.g., thermistors or other temperature sensors), one or more heating elements 405, and temperature control circuitry 407 for controlling the operation of the heating elements (e.g., powering the heating elements as needed to reach a predetermined or programmed temperature setpoint or temperature range). FIG. 5B illustrates an example of this temperature compensation operation, showing a time-varying heater output generated according to a time-varying ambient temperature to maintain a constant or near-constant resonator temperature.

[0043] As FIG. 5A illustrates, the integration of the DDS resonator and temperature compensation system components can vary from implementation to implementation. For example, in one embodiment, shown within shaded region 411, DDS resonator 401 is integrated (i.e., fabricated or otherwise formed together on the same substrate or die) with temperature sensor 403 (“Integrated T-Sense”), while temperature control circuit 407 and heater 405 (i.e., one or more heating elements) are implemented external to the die. In another embodiment, shown by region 415, temperature control circuit 407 and heater 405 are integrated with temperature sensor 403 and DDS resonator 401, thereby establishing a fully integrated active temperature compensation system within the MEMS resonator die. Although not specifically shown, DDS resonator 401 may alternatively be integrated with heater 405, with one or more temperature sensors and / or temperature control circuitry remaining external to the die.

[0044] In some embodiments, the temperature sensor shown in Figures 4A and 4E may be implemented within one or more component parts of an encapsulated dual silicon resonator, or within a multi-die package including a dual silicon resonator (MEMS) die and a logic die. Figure 6 illustrates various options for such temperature sensor integration, including temperature sensor implementation within the lid or device layer of the encapsulated dual silicon resonator and / or within a co-located logic die that also houses sustain / sense circuitry for driving the dual silicon resonator into one or more resonant oscillation modes and sensing resonant mechanical motion of the resonator. In the case of in situ placement within the device layer of the encapsulated resonator structure, the temperature sensor may be implemented by a second micromachined structure—more specifically, a micromachined (or MEMS) thermistor structure having any number of geometries that are substantially free from the device layer substrate and allow relatively stress / strain-free thermal expansion and contraction (and therefore limited mechanically induced nonlinearity in the temperature dependence of the end-to-end thermistor resistance). Although not specifically shown, the end terminals of the MEMS thermistor may be bonded to TSVs in the encapsulation structure (lid and / or substrate) and ultimately to package-level contacts to enable a temperature-sensing output signal. When implemented in the lid wafer or logic die, the temperature sensor may be implemented with silicon, polysilicon, or metal thermistor features, p-n junctions, and / or other thermally sensitive features. In yet other embodiments, described in more detail below, the temperature sensor may be implemented with patterned traces and / or p-n junctions in one or more layers of the dual silicon resonator itself.

[0045] 7 illustrates an exemplary process for fabricating a MEMS thermistor and a dual silicon MEMS resonator (having the two degenerately doped silicon and piezoelectric layers described above) in adjacent regions of the same SOI substrate (i.e., within the same encapsulated device layer). As shown, an oxide layer is deposited on the SOI substrate and then etched (e.g., wet etched) to provide a masked thermistor region. After PSG deposition (e.g., via vapor-phase POCI3) and a mini-drive cycle (a relatively short duration drive) to achieve a relatively shallow dopant diffusion, a long anneal (a relatively long and / or high temperature drive) is then performed to achieve a deeper final dopant diffusion profile. At this point, the single-crystal silicon device layer is degenerately doped (with a process-dependent gradient) except in the thermistor region below the oxide mask, which remains relatively undoped. The remaining layers of the dual silicon resonator (e.g., AlN and ISDP, or any other material according to the above-mentioned variations) are deposited, the oxide mask is removed, and a trench / etch and release process is performed to complete the two co-located MEMS elements—the dual silicon resonator and the adjacent MEMS thermistor. By juxtaposing the MEMS thermistor and dual silicon resonator within the same device layer and within the same (eventually) encapsulated chamber and singulated die, the temperature gradient between the thermistor and resonator that is problematic in temperature sensing / resonator configurations that are not relatively close together is eliminated (or made negligible), thereby improving the accuracy of the various temperature compensation schemes included in FIGS. 4A and 4E.

[0046] 8 shows an example interconnection configuration within a dual silicon resonator with an additional lightly doped (or undoped) single-crystal silicon thermistor layer as part of the resonator structure. While this configuration adds a layer to the resonator material stack (in this case, the thermistor layer is disposed below the dual degenerately doped silicon and piezoelectric layers—although any other stack location may be feasible), thermal coupling of the thermistor layer to the remainder of the resonator material stack ensures a zero or near-zero temperature difference between the two resonator regions, again avoiding error-inducing temperature gradients that are problematic in less closely spaced configurations. As shown, an additional pair of sense terminals may be coupled across the thermistor layer (e.g., through conductive structures within the tethering / tethering structures to opposite lateral ends of the thermistor layer) and routed to external contacts of the encapsulated material stack and final resonator / thermistor package.

[0047] 9A illustrates one embodiment of a DDS resonator 500 having some of the above-described features. More specifically, the DDS resonator 500 may be a single-layer structure (i.e., composed of only a single degenerately doped semiconductor, such as degenerately doped silicon), or may be a multi-layer structure with, for example, an aluminum nitride (AlN) or other piezoelectric layer disposed between the degenerately doped silicon layer and a conductive layer, as shown in FIGS. 1A-1C. Also, the dopant type or concentration within the DDS layer or any other layer of the resonator 500 may be non-uniform (e.g., the concentration may be relatively high or low in the high-stress area 501 or other areas of the resonator body), and the resonator may be fabricated such that the resonator axis 503 is disposed at a non-zero angle φ (i.e., the "resonator angle") with respect to the crystal axis 504. As described above, the resonator angle, DDS dopant concentration, and type (including any non-uniformities) may be specifically engineered to null the first-order and at least one higher-order frequency temperature coefficient, respectively. The mode shapes, relative layer thicknesses, and dopant type / concentration of other layers (e.g., conductive layers formed from degenerately doped polysilicon) may also be specifically selected, in combination with the dopant concentration / type of the bulk DDS layers, to provide desired temperature-dependent resonant operation, such as temperature-indicating operation or temperature-stable (or temperature-insensitive) operation, over one or more desired operating temperature ranges. In the particular example shown, DDS resonator 500 has an elliptical shape with a narrowed transverse dimension (i.e., perpendicular to resonator axis 503) between spring bearing anchors 507a and 507b. Trenches 512 are etched or otherwise formed around and / or beneath the DDS resonator and anchor spring elements to free them from the substrate field area 510 (i.e., to free these elements to move relative thereto).

[0048] The DDS resonator 500 can have the AA cross section shown schematically in Figure 1B—i.e., a degenerately doped (DD) single crystal silicon layer, an aluminum nitride piezoelectric layer, and a degenerately doped polysilicon electrode layer. Spring elements ("springs") and field area anchors are disposed on opposite sides of the resonator body to form the individual anchors shown at 507a and 507b in Figure 9A. In alternative embodiments, including single anchor configurations or configurations with more than two anchors, various other tethering configurations, with or without spring mounts, can be utilized.

[0049] Figure 9B shows an example finite element model of the DDS resonator of Figure 9A, illustrating displacement and stress distribution during resonant vibration (or oscillation). For example, vectors (arrows) protruding from the edges of the resonator body indicate the direction of resonator movement during the expansion phase of the oscillation cycle. In the illustrated example, high stress areas occur between the anchors, low stress areas occur at the opposite ends of the resonator axis (i.e., the axis shown in Figure 9A), and stress contours occur between the low and high stress areas. Stress in the anchors is approximately zero.

[0050] 10A-10I show non-exhaustive examples of alternative resonator geometries / designs, with each example resonator implemented in a 0110 crystal orientation (i.e., an axis offset by 45 degrees from the monocrystalline silicon layer axis and / or the major axis of the polysilicon layer, although various other angular orientations may be selected according to TCF industrial design objectives). Referring first to the resonator shown in FIG. 10A, centrally located tethers (springs) on either side secure the otherwise free resonator body to field area anchors. Increased resonator end mass tends to confine (limit) mode interaction outside of the desired resonant frequency over the frequency range of interest.

[0051] 10A-10I, resonant motion generates selected directional stresses across the resonator body in terms of the in-plane anisotropic single-crystalline silicon core, particularly in the illustrated 0110 axial orientation. Slots or vents (e.g., approximately 7 microns wide) are provided in the resonator body to provide stress relief and facilitate gas-phase HF evacuation, and also to enable lateral dopant diffusion in at least some processes (e.g., as described above with reference to FIGS. 2B and 3I-3K) and / or to facilitate industrial design of TCFs. For example, in the resonators of FIGS. 10B and 10C (and others), the location and size of the slots are selected to enhance dopant concentration in regions of high stress and, conversely, to reduce dopant concentration in regions of low stress. In the resonator designs of Figures 10D, 10E, 10F, and 10G, vents in the field area (outside the resonator body) may be used to release these regions during gas-phase HF etching, thereby forming cantilevers or other structures that suppress / avoid field-area interference with resonant operation. The example design in Figure 10G includes a single, centrally disposed anchor / tether from which the release vent extends and which is tuned to limit transmission of vibrations to the resonator bulk. The embodiments of Figures 10H and 10I show a centrally tethered resonator with more than two released members (or appendages) extending from it. The resonator in Figure 10H, for example, resonates primarily in extensional mode (each of the point masses moves in-plane with respect to the anchor), while the four beam members extending outward from the central anchor of the Figure 10I embodiment resonate in extensional mode (effectively acting like a tuning fork with identical ends). In alternative embodiments, including embodiments that intentionally exhibit multiple simultaneous resonant modes (e.g., with different TCF characteristics, as described below), various other resonator core shapes, tethering schemes, number of appendages, angular orientations, etc. may be implemented.

[0052] Although omitted in some dual silicon resonator embodiments described above, one or more metal layers or silicides may be introduced into such structures to improve electrode conductivity and / or to act as a seed bed for improving the material structuring of the piezoelectric layer (e.g., growing relatively regular vertical grains in an aluminum nitride piezoelectric layer). FIG. 11A illustrates a scheme involving a layer such as titanium nitride or molybdenum nitride formed on degenerately doped single-crystal silicon prior to piezoelectric layer formation, with the top electrode again implemented by a layer of heavily doped polysilicon. FIG. 11B illustrates a similar scheme in which metal deposition (e.g., titanium or tungsten) on a degenerately doped single-crystal silicon layer is thermally driven to form a silicide seed bed (e.g., TiSi2 or WSi) for improved piezoelectric layer formation and bottom electrode conductivity. Although not specifically shown, the layer stack illustrated in FIG. 11A may be reordered by reversing the positions of the degenerately doped poly and metal layers. Such a configuration may be more susceptible to aging / strain hardening, but may improve the conductivity of the top electrode without sacrificing the thermal frequency trim authority provided by the degenerately doped polysilicon layer.

[0053] 12A and 12B illustrate an alternative resonator material stack having the three layers shown in FIG. 1A (i.e., two degenerately doped silicon layers and one piezoelectric layer) along with an additional metal layer. More specifically, in FIG. 12A, the additional metal layer is disposed between the degenerately doped poly layer and the piezoelectric layer and functions, at least in part, to improve the conductivity of the top electrode. In one embodiment, for example, the metal layer functions solely as the top electrode for the piezoelectric layer (e.g., with a field area via direct contact to the buried metal layer), and in this case, the degenerately doped poly layer can function primarily as the thermal frequency trim authority. In the embodiment of FIG. 12B, the degenerately doped poly layer is relocated to the opposite side of the piezoelectric layer (i.e., sandwiched between the piezoelectric layer and the core DD single crystal silicon layer), thereby improving the conductivity of the bottom electrode and maintaining the thermal trim authority of the poly layer. Also, although not specifically shown, an additional metal layer may be disposed adjacent to the lower surface of the piezoelectric layer (i.e., two metal layers sandwich the piezoelectric layer) to minimize conductive losses without sacrificing the TCF industrial design authority provided by the degenerately doped monocrystalline silicon core layer or the thermal trim authority provided by the degenerately doped poly layer. Furthermore, while dual degenerately doped silicon layers (monocrystalline and polycrystalline silicon) are components of all material stacks described thus far, in all cases (including the embodiments described below), either or both of these layers may be lightly doped, undoped, or in some cases omitted entirely, e.g., in which case their contribution to the resonator bulk, TCF industrial design, thermal frequency trim, and / or other industrial design or device finish authority is unnecessary.

[0054] As mentioned above, the layer thicknesses in a dual silicon resonator are typically non-uniform and are selected to provide, for example, desirable TCF characteristics (e.g., the core single crystal silicon layer is 30x the width of the aluminum nitride piezoelectric layer) as well as the size, aspect ratio, or other dimensions of the bulk resonator. However, due to their different surface tensions, the material layers tend to warp when laminated into a composite stack, complicating downstream manufacturing steps (and possibly producing undesirable modal characteristics or otherwise degrading runtime performance). Figures 13A and 13B show an embodiment of a symmetric dual silicon material stack that avoids the surface tension imbalance (and resulting warpage). In the embodiment of FIG. 13A , for example, after fabricating two dual silicon / piezoelectric layer material stacks generally as shown, one of the material stacks is inverted and bonded to the other to result in a six-layer stack (or five-layer stack, if the bonded polysilicon layers are considered a single layer) with lines of symmetry at the bonding surface. Thus, a relatively thick degenerately doped single-crystal silicon core at the bottom of the stack is balanced by a degenerately doped single-crystal silicon core of the same height (“h1”) at the top of the stack. Electrically, the stack can be viewed as having top and bottom electrodes (formed by matching degenerately doped single-crystal silicon layers) with series-connected capacitors (two piezoelectric material layers) coupled together through a degenerately doped polysilicon layer bonded and bonded between the respective electrodes—a pair of series-coupled capacitors electrically equivalent to a single capacitive element with twice the elementary capacitance. One advantage of this approach, in addition to balancing surface tension forces, is the relatively high material stack profile that results from the two single crystal silicon (core resonator) layers, a configuration that can meet large profile design specifications (e.g., resonator heights of 12-20 microns or more) without requiring excessive process scaling or additional material layers.

[0055] Figure 13B shows an alternative symmetric dual silicon material stack in which a layer of bondable conductive material is formed between the core single crystal silicon layer and the piezoelectric layer, after which another such layer (sandwiching the piezoelectric layer between two thin films of bondable conductive material) is formed on the piezoelectric layer. After surface preparation (e.g., chemical-mechanical planarization), another of the core silicon layers is bonded to the material stack, thereby forming a substantially symmetric material stack (minus the lower insulator and handle wafer) with the same series-coupled capacitor electrical model as the material stack of Figure 12A. In a variation of the embodiment of Figure 12B, the piezoelectric layer may be sandwiched between non-conductive bonding layers.

[0056] The following are at least some of the advantages that may be realized by the above-disclosed embodiments. Degenerately doped silicon can replace two separate materials used in conventional piezoelectric resonators: a temperature-compensating material (e.g., SiO2) and a conductive material (e.g., Mo). Degenerately doped silicon has the ability to compensate temperature and also provides a sufficiently low electrical resistance (e.g., 1-50 ohms) to function as an electrode material for many applications. Degenerately doped single-crystal silicon facilitates improved frequency stability with temperature by allowing both first- and second-order TCF nulling. The doping and orientation of the degenerately doped silicon layer provide at least two degrees of freedom for nulling at least two temperature coefficients. The above examples demonstrate that first- and second-order TCF nulling of composite piezoelectric resonators utilizing heavily doped silicon is possible and within an accessible design space. The temperature coefficient improvements enabled by the disclosed embodiments result in a potential frequency variation of less than + / - 10 ppm across the industrial temperature range of -40 to +85°C. In contrast, micromechanical resonators with only first-order TCF compensation typically exhibit a variation of + / - 50 to 200 ppm across the industrial temperature range. The removal of oxide and metal interfaces improves the mechanical quality factor of the resonator through the removal of potentially lossy layers and interfaces that can significantly increase the mechanical losses of the resonator. Furthermore, the replacement of metal and oxide layers in the structure with semiconductors eliminates work hardening, fatigue effects, and interface effects that contribute to frequency hysteresis with temperature and frequency aging with time. Resonators with engineered TCF characteristics can be created from one or more layers of material. Within a layer, first- and second-order TCF coefficients can be nullified through design and doping, provided the material has qualities similar to degenerately doped silicon. Within a layer with one or more material properties, there may be one or more regions. For example, a single silicon layer may have at least one region containing at least one doping level and at least one dopant. A single material layer may have at least one region containing at least one material type. Each of these regions may add an additional degree of freedom in the TCF behavior of the resonator. Adjusting the size and properties of these regions can affect the nominal frequency and first-, second-, third-, and higher-order TCF terms.

[0057] Micromachined structures in microelectromechanical systems (MEMS) may benefit from in-fabrication or post-fabrication heating for a variety of reasons. While traditional in-fabrication heating in an oven or furnace can be effective when a common temperature needs to be achieved across an entire MEMS wafer or batch of wafers, new fabrication and post-fabrication processes often require repeated heating at precise die-specific or wafer-specific temperatures. In addition, some processes and / or products may benefit from non-uniform (or regional or localized) heating that is not easily achievable by whole-die or whole-wafer oven processing.

[0058] In addition to the various temperature-stable and wear-resistant properties described above, the degenerately doped polysilicon and / or degenerately doped single-crystalline silicon layers of the resonator structures described above may be thermally cycled via Joule or radiative heating techniques to achieve targeted post-encapsulation and / or post-packaging tuning of the resonator frequency—e.g., tuning the natural frequency of a wafer-scale or singulated resonator to within 50 ppm, 20 ppm, 10 ppm, or less of a target output frequency.) More generally, hereinafter, on-die structures that enable heating or superheating of micromachined elements within MEMS devices (i.e., including, without limitation, the resonator structures described above) are disclosed in various embodiments. In a first group of “contact heating” embodiments, a direct current conduction path is established through the target micromachined structure to generate ohmic or resistive heating (“Joule heating”), which in some cases allows the target element to be heated (i.e., “superheated”) to temperatures significantly exceeding the nominal MEMS operating temperature range. In some contact heating embodiments, a "tether" structure is used to interconnect the target micromachined elements to individual anchors to source and sink DC current and, optionally, to provide thermal insulation relative to the anchors and device substrate. In other contact heating embodiments, one or more electrode structures formed within the material stack of a MEMS resonator (or other movable micromachined member) are patterned to create one or more device heating hot spots.

[0059] In a group of "capacitively coupled heating" embodiments, an alternating current (AC) flow is generated within the target micromachined element through application of a sufficiently high frequency signal to an adjacent electrode (i.e., a non-movable electrode that is capacitively coupled to the movable micromachined member and, in the case of a micromachined resonant member, may function in device operation as a resonator drive electrode and / or resonator sense electrode). In still other embodiments, direct or capacitive coupling to individual dielectrically separated layers of the micromachined structure is utilized to transport radio frequency (RF) energy across the dielectric boundary and thereby heat the separated layer. In all such embodiments, the MEMS resonator structure or portions thereof (e.g., one or more layers of a multi-layer device) may be patterned to improve or delay Joule heating effects. Also, in all cases, multiple energy delivery ports may be provided to enable localized heating of individual regions and / or layers of the micromachined member. In self-heating embodiments, programmable steering circuitry (to steer AC or DC current to a selected energy supply port or group of ports) and / or programmable heating control circuitry (to establish a desired heating profile) may be provided on the same die as the target micromachined member or on a separate die in a multi-die package. In still other embodiments, radiative heating (e.g., laser or radio frequency heating) may be used to heat the MEMS resonator structure or portions thereof, including, for example, transmitting radiant energy through a transparent (i.e., transparent to the relevant wavelengths) lid or other structure of an encapsulated or packaged MEMS resonator. These and other embodiments and features are disclosed in further detail below.

[0060] 14 illustrates an exemplary micromachined member (or structure) within a microelectromechanical system (MEMS) that can be heated or superheated using the various techniques and circuit configurations disclosed herein. In the configurations shown and embodiments described below, the micromachined member is a movable resonant member within a MEMS device and includes at least one degenerately doped semiconductor (DDS) layer in accordance with various embodiments described above. In all cases, the micromachined member may alternatively be a movable mass within an accelerometer, thermistor, or any other useful MEMS structure and may lack a DDS layer. Also, the DDS layer, if present, may have a density of, for example, 1e18 atoms / cm. 3 , 1e19 atoms / cm 3 , 1e20 atoms / cm 3 , or 2e20 atoms / cm 3 It may be implemented by doping single crystalline, polycrystalline, and / or amorphous forms of silicon, germanium, etc., with relatively high concentrations of dopants (e.g., arsenic, boron, phosphorus, antimony, etc.), such as greater than 1000 . Accordingly, "DDS" should be understood to refer not only to degenerately doped silicon, but also to various other degenerately doped semiconductors, including, for example, and without limitation, degenerately doped germanium, 3-5 compounds, silicon-germanium, etc. Although degenerately doped silicon is referenced in some instances herein, it is solely because it is the most common semiconductor in the industry, and such reference should not be considered limiting.

[0061] 14, the resonant member may be secured to one or more anchors by respective tethering structures (“tethers”) and disposed between the respective drive / sense electrodes, and / or may have electrodes patterned on top of or within an intermediate layer. Detail views 615, 617, and 619 show several possible implementations of the resonant member, including a resonant structure formed entirely from DDS (e.g., degenerately doped silicon), a DDS layer having a layer of piezoelectric material disposed on top of it, and a piezoelectric layer sandwiched between the DDS layer and the electrode layer. In the latter three-layer implementation, the electrode layer may include one or more conductive electrode regions separated from each other by a dielectric (as shown) or merged into a unified electrode or electrode layer.

[0062] 15A shows a profile view of an example MEMS device having a DDS resonator (an example of a movable micromachined or microelectromechanical component) along with structural interconnections that allow Joule heating during fabrication. In the illustrated contact heating embodiment, the DDS resonator is secured between anchors by a multi-function tether. That is, the tether not only mechanically secures the DDS resonator to the anchors (and thereby to the device substrate), but also controls the Joule heating current (I JH ), and in some implementations, as thermal insulation between the DDS resonator and the anchor.

[0063] The anchors themselves may be implemented by or layered with a conductive material and thus be capable of conducting Joule heating current from the wafer probe or singulated die probe. More specifically, a voltage source (or current source) established between the two probes (i.e., the potential in either case indicated by v+ and v-) prior to device encapsulation generates a Joule heating current that flows from left to right in the illustrated example, through the leftmost anchor and tether, through the DDS resonator, and then through the rightmost tether and anchor, thereby Joule heating the tether and DDS resonator. Also, although DC current is shown in Figure 15A (and other following figures), AC current may also be conducted through the DDS resonator.

[0064] FIG. 15B shows a Joule heating MEMS embodiment after encapsulation, in which conductive vias extend through the hermetic encapsulation of the lid configuration to make electrical contact with the conductive tethering structure (i.e., the anchors are formed from or layered with a conductive material, similar to FIG. 15A). With this configuration, Joule heating current may be driven through the DDS resonator in a finishing operation after encapsulation, or possibly even after fabrication, including Joule heating in the field before or after the MEMS device is deployed in a host system. FIG. 15C shows an alternative embodiment in which conductive vias extend through the device substrate (e.g., bulk semiconductor) to enable Joule heating current flow through the DDS resonator. Note that in both the FIG. 15B and FIG. 15C embodiments, the hermetic encapsulation, along with a peripheral sealing ring or similar structure, encloses a vacuum cavity or a cavity filled with an inert, insulating material, thereby limiting heat radiation from the DDS resonator to the substrate and lid.

[0065] 16A-16C illustrate embodiments of multifunctional tether structures, including examples of bulk tethers (FIG. 16A), composite tethers (FIG. 16B), and extended path tethers (FIG. 16C). In general, the thermal and electrical resistance of a given tether structure is inversely proportional to its cross-sectional area, and the mechanical stiffness imparted by the tether (i.e., for securing the movable micromachined member relative to the anchor) increases with cross-sectional area. In the bulk tether embodiment of FIG. 16A, the width of the tether (W—a proxy for cross-sectional area in embodiments with a depth dimension fixed according to the depth of the micromachined member) is selected according to the thermal and electrical conductivity of the tether material and the distance it extends (D) to provide the desired mechanical stiffness as well as the desired electrical and thermal resistance. For example, the electrical and thermal resistance of the tether may be engineered to achieve a desired temperature range within the tether itself, thereby providing a thermal barrier (insulation) between a heated or overheated movable micromachined member and an ambient (or near ambient) temperature anchor and substrate.

[0066] In the embodiment of FIG. 16B , the “composite” tether is implemented by an assembly of N component tethers (N>1), each sized to provide a desired cross-sectional area in terms of extension distance and to provide a predetermined degree of thermal insulation and mechanical stiffness. Depending on the intended motion of the movable micromachined member (e.g., different possible vibration modes within the resonant member), decomposing the bulk tether into multiple component tethers shown in FIG. 16B may provide improved freedom of movement without unduly compromising mechanical support. Decomposing the bulk tether into multiple component tethers that collectively occupy the same footprint as the bulk tether reduces the cross-sectional area of the composite tether (compared to a comparable bulk tether), thereby increasing electrical and thermal resistance values so that, for the same Joule heating current and fabrication materials, higher tether temperatures can be achieved—a result that may enable higher temperatures within the target micromachined member.

[0067] In the embodiment of FIG. 16C, the tether structure is routed laterally and folded upon itself in a direction orthogonal (perpendicular) to the extending distance, such that electrical and thermal paths are extended between the anchor and the movable micromachined member, so that significantly greater temperature and potentially greater thermal insulation can be achieved. In the particular implementation shown, the tether includes symmetric lateral extensions or folds (i.e., extending a lateral distance L in both directions in relation to the axis of extension). With this configuration, and by selecting a collective lateral extension distance (2*L, where "*" denotes multiplication) that approximately matches the width of the movable micromachined member (or at least the width of the anchor), the tether presents an effective thermal barrier between the movable micromachined member and the anchor. The lateral extension dimension of the folded tether may, in an alternative embodiment, be only a portion of the width of the movable member (e.g., 1 / 4 or 1 / 3 of the width of the movable member, although smaller or larger portions can be implemented), in which case it should be noted that multiple composite folded tethers can be deployed. Also, as described above, the width of the tether (W, which may be uniform throughout the base and lateral extensions of the tether, or relatively narrow on one or the other) can be industrially designed to provide desired thermal and electrical resistance values, and thereby generate a relatively high temperature within the tether, resulting in an increased effectiveness of the tether as a thermal insulator (e.g., the ratio of L to W can be set to 2, 3, 4, 5, or more than these, or any fractional value between these integer ratios). Furthermore, the U-shaped shape of each lateral (folded) extension provides a degree of mechanical compliance (yet another function performed by the tether), thereby compressing and expanding with the movement of the micromachined member.Thus, the tether shown in FIG. 16C functions to mechanically secure a movable micromachined member to an anchor, provide an electrical path for Joule heating, thermally isolate the movable micromachined member from the anchor, and provide spring-like mechanical compliance between the micromachined member and the anchor. In an alternative embodiment, the lateral extension dimension L (which may be smaller than the extension distance D, or approximately or nominally equal to it, or larger than it) may be larger or smaller than that shown (e.g., it may extend beyond the contour of the movable micromachined member), and the shape of the lateral extension may be different (e.g., it may curve radially in a semi-circle or semi-ellipse instead of having bends / folds to form a U shape). Also, instead of a symmetric lateral extension, the tether can have a single lateral extension (i.e., in only one direction in relation to the extension axis). Similarly, instead of a single tether having symmetric lateral extensions and folds, the MEMS device may include two single-folded tethers having oppositely directed lateral extensions (i.e., as if the tether shown were split into two along the extension axis), or, in some cases, it can include multiple pairs of such folded tethers where the fold of a given tether is included within the fold of another.

[0068] 16D shows an example thermal profile achieved by conducting DC or AC Joule heating current through the movable micro-machined member via the anchor and folded tether. Due to the relatively high electrical resistivity of the tether in relation to the anchor, and in at least some cases, to the movable micro-machined member, Joule heating occurs primarily within the tether itself, resulting in conductive and possibly radiative heating of the movable micro-machined member (in which case Joule heating may also occur). The tether also insulates the movable micro-machined member from the anchor, which is at or near ambient temperature (T0), during Joule heating, so that in a vacuum environment, at least the nominal tether temperature at the physical midpoint of the tether (i.e., between the anchor and the micro-machined member) is T0 + T0. JH / 2, and the movable micromachined component reaches a temperature of T0 + T JH The temperature profile will be relatively uniform throughout. In alternative embodiments, other temperature profiles (including those with non-linear gradients) may be generated. The movable micro-machined member may be slightly cooler than the hottest part of the tether due to radiation losses or gas conduction, but will typically be near the highest temperature of the tether. The temperature profile can be taken into account in the design and trim process.

[0069] 17 shows another Joule-heating MEMS embodiment in which individual pairs of "flexible" tethers are provided to conduct Joule heating current to a movable micro-machined member, with each such tether pair effectively forming a dedicated heating port. In this configuration, the mechanical tethers function to connect the micro-machined member to anchors (two are shown, although a single-anchor configuration is also possible) and provide the desired mechanical stiffness, but the flexible tethers can also be designed to be relatively "soft," or in some cases as soft as possible, to minimize damping of vibration or inertial motion of the micro-machined member. The provision of multiple flexible tether pairs also allows for non-uniform heating or heating of individual regions of the micro-machined member, an effect that may be useful in some applications. In alternative embodiments, more or fewer heating ports (flexible tether pairs) may be provided, including heating ports that drive current across corners, individual layers, or other more isolated regions of the micromachined member.

[0070] 18A and 18B illustrate an exemplary capacitively coupled actuated heating configuration in which an electrode functioning as a drive and / or sense electrode (e.g., in a MEMS resonator or accelerometer) also serves as a radio frequency (RF) energy input that drives AC Joule heating current in a movable micro-machined member. In the embodiment of FIG. 18A , for example, an RF energy source is supplied to both electrodes, and complementary phases of the RF signal (or, for example, a steady-state neutral voltage set midway between the upper and lower peaks of the RF signal) are coupled to one or more multifunction tethers, thereby enabling AC current flow to and from the movable micro-machined member via the multifunction tether. That is, energy is transferred from the RF energy source to the micro-machined member via capacitive coupling between the electrode and the micro-machined member, with charge carriers (i.e., AC current) alternately being released and extracted from and to the micro-machined member via the multifunction tether.

[0071] In the embodiment of Figure 18B, complementary phases of an RF energy source (or an RF signal and a neutral voltage) are coupled to two electrodes in an alternating manner, causing the electrodes to be differentially driven to generate AC current flow therebetween (i.e., charge carriers are forced across the movable micro-machined member in an alternating manner toward one electrode and then toward the other). In such an embodiment, the mechanical tether (or tethers) may be non-conductive (if conductive and coupled to a fixed potential substrate via conductive anchors, AC current would flow similarly to Figure 18A), and are generally engineered as described above to provide thermal insulation relative to one or more anchor structures that have a temperature at or near the ambient temperature surrounding the die or wafer.

[0072] 19A shows yet another Joule heating embodiment in which energy is supplied to individual layers of a multilayer device either conductively (e.g., via the conductive tethers described above) or capacitively (e.g., via the capacitive electrodes described with reference to FIGS. 18A and 18B). In the particular example shown, the top electrode layer and the bottom DDS layer sandwich inner dielectric layers (e.g., piezoelectric material) such that the outer layers can be thought of as forming individual resistances through which DC or AC Joule heating currents are passed.

[0073] FIG. 19B shows an alternative heating configuration, in which Joule heating is primarily provided within the patterned electrode layers of a dual silicon resonator (i.e., a degenerately doped poly layer and a degenerately doped single crystal silicon layer with a piezoelectric layer sandwiched between the silicon layers). The tether may still be structured to provide thermal insulation between the resonator and the tethering field area, but is otherwise sufficiently conductive so that the primary Joule heating voltage drop (and power dissipation) occurs within the patterned electrode layers. In the particular example shown, the patterned electrodes (e.g., implemented in one or both of the degenerately doped silicon layers—single crystal silicon and / or polysilicon, the latter configuration being shown) comprise relatively narrow conductive passages extending between relatively large bulk interconnect nodes. Due to their relatively small cross-sectional area (and correspondingly relatively large resistance), the narrow passages provide ohmic (I) conduction within the conductive path traversed by the Joule heating current and, therefore, the primary Joule heating source. 2 R) constitutes the primary source of power dissipation. Similar to the tether-based heating configuration of FIG. 16D, the resonator maintains a target temperature (e.g., T + T) by limited thermal radiation from the resonator to the substrate and lid structure. JH) can be heated relatively uniformly. More generally, patterned electrode Joule heating, alone or in combination with other Joule or radiant heating sources, can produce temperature profiles that differ from the example of FIG. 16D, including producing localized or non-uniform temperature distributions.

[0074] Figure 20 shows another embodiment in which an RF energy source is coupled across the outer (at least moderately conductive) layers of the movable micro-machined member shown in Figure 19 A. With this configuration, AC currents are generated in the electrodes and DDS layers of the micro-machined member to allow heating or superheating of those layers.

[0075] FIG. 21 illustrates an encapsulated or chip-scale MEMS package having a MEMS die (e.g., having movable micromachined members formed thereon) and a control die. As illustrated, the control die includes not only several low-power control signal interconnects that carry relatively low-power runtime control and output signals between the two dies, but also relatively high-power thermal energy interconnects that enable the supply of Joule heating energy in the form of DC or AC power signals. While a single pair of thermal energy interconnects is shown, more interconnects can be provided in embodiments having multiple heating ports. Also, while the thermal energy interconnects are shown extending between the control die and the MEMS die, they may alternatively or additionally be exposed at the surface of the package to enable power supply from an external heating controller. Furthermore, the thermal energy interconnects between the MEMS die and the control die may be omitted due to the energy input interface on the MEMS die being accessed before or after encapsulation, as described above (thus allowing MEMS die heating to be performed prior to packaging with the control die). Electrical vias and contacts sized for overheating may be larger in lateral dimensions than those designed for normal operation. For example, electrical vias for normal operation may be 5-15 microns in diameter to minimize parasitic capacitance. In contrast, vias sized for overheating may have diameters ranging from 20-100 microns to minimize resistive heating within the vias, for example and without limitation. The vias may be in contact with a low melting temperature material (such as aluminum or titanium), and thus, the increase in via temperature will be limited. Furthermore, the conductive material within the vias may not be a lightly doped semiconductor, but may have a lower electrical resistivity than would otherwise be required for normal operation, such as a DDS.

[0076] 21 , a programmable heating control circuit may be provided in the control die to enable in-situ post-fabrication heating or heating of one or more micro-machined features within the MEMS die. In one such embodiment, the control die may also include a programming interface (i.e., signaling interface) accessible via external contacts of the MEMS package. Thus, an external controller (e.g., a dedicated and in-situ controller with the MEMS package within a host system, or a stand-alone manufacturing or post-fabrication controller) may be coupled to the programming interface to issue the programming commands and operands needed to configure the programming heating control circuit and thereby initiate and control the heating / heating of one or more micro-machined features within the MEMS die.

[0077] Figure 22 illustrates one embodiment of a programmable heating controller that may be used to implement the programmable heating control circuit of Figure 21. As shown, the heating controller includes several (N) programmable register banks that provide individual sets of programmed parameters to a power driver circuit. The power driver circuit then drives a power signal (i.e., functions as a voltage or current source) to each individual heating port according to the parameters programmed in the corresponding register bank. In the particular example illustrated, each programmable register bank includes a port enable field that allows selective enabling or disabling of the heating port, as well as a set of register fields (any or all of which may be implemented by individual registers) for storing individual parameters that control the amplitude, duration, and frequency of the voltage or current pulses output by the power driver through the corresponding heating port.

[0078] 23 illustrates a configuration in which multiple heating power drivers within an external heating controller may be coupled to individual singulated or on-wafer MEMS devices, thereby enabling the simultaneous performance of the various heating and overheating operations described above in connection with the subject MEMS devices. The external heating controller may be implemented, for example, within manufacturing or test equipment (e.g., within automated test equipment (ATE)) that may perform numerous other functions related to or independent of the heating / overheating of the MEMS devices.

[0079] Figure 24A shows an example heating power profile that may be generated within the external or in-situ heating controller of Figures 22 and 23, along with an example temperature generated within a movable micromachined member (i.e., within a MEMS device or package). In the particular embodiment shown, pulses of various widths (durations) and amplitudes (powers) are generated, with the net energy delivery being approximately proportional to the area under the pulse. More specifically, pulse durations may be varied from at least 0.1 ms to 100 ms (alternative embodiments may implement shorter and / or longer pulse durations), and power levels may range from at least 20 mW to 500 mW (higher and / or lower limits may be supported). During heating / superheating operations, temperatures generated within the movable micromachined elements may significantly exceed nominal MEMS runtime operating temperatures, which in the illustrated example thermal profile have a range of 300-1300°C, although higher and / or lower temperatures (e.g., lower limits of 500, 600, 700, or 800°C) may also be achieved.

[0080] Also, although not shown in detail in FIG. 24A , the pulse frequency may be varied in accordance with heating control parameters (e.g., as described with reference to FIG. 22 ). Also, as shown in FIG. 24B , the power pulse waveform may be shaped to achieve a controlled cooling profile, for example, by selecting between various power-down ramp rates. Thus, if a relatively slow cooling rate is desired, a relatively long (relatively flat slope) power-down ramp may be selected, and conversely, if a relatively fast cooling rate is desired, a relatively short (relatively sharp slope) power-down ramp may be selected. The selected power-down ramp rate may be applied to all power pulses (in which case different ramp rates may be applied to individual pulses) or may be applied only to the final pulse or pulses in a heating (power pulse) sequence. Furthermore, while linear power-down ramps are shown, any or all of the ramps may have nonlinear profiles (e.g., exponential or hyperbolic profiles). Also, although a controlled power down profile is shown, the rising edge of the power pulse may be correspondingly shaped to achieve a controlled heating rate as well. In all cases, power profiles other than (or in addition to) the pulsed profiles shown may be used to achieve device heating operation, including, for example and without limitation, steady-state or non-return-to-zero power profiles.

[0081] The electrical circuitry used to heat the structure can also sense the temperature of the structure. The electrical properties of the structure are generally a function of temperature; for example, resistance may increase with temperature. This may be used to measure the temperature before or after trimming. It may also be used to measure the temperature of the structure between heating pulses, or even during the heating pulse. Thus, pulse duration or height or other power profile characteristics may be varied beforehand or as the tether progresses to achieve optimal results. For example, if the tether has a particular resistance when cold and a different resistance when it is at a desired high temperature, current pulses may be applied until the desired high temperature is reached. The resistance of the structure can be simultaneously measured as it heats by various means, such as by applying a current to heat the structure while measuring a voltage across the structure to determine its resistance, or alternatively, by applying a DC heating drive while measuring resistance with an AC test signal.

[0082] 25 illustrates exemplary processing of a packaged MEMS device in which an oven reflow operation is performed to merge two separate terminals that form a heating port into a single operational terminal. That is, prior to oven reflow, the packaged MEMS device includes three electrically isolated terminals, two of which form ports for heating or superheating internal micromachined structures as described above, and the other forms the first operational terminal. As the individual solder layers (or layers of other heat-fusible conductive material) of the two heating port terminals become fluid and flow together during oven reflow, the two heating port terminals effectively merge to form the second of the two operational terminals (i.e., the MEMS device is converted from a three-terminal device to a two-terminal device). This same approach may be performed with respect to the first operational terminal (i.e., reflow to merge separate terminals that may form a second heating port into the first operational terminal) or with respect to other terminals not shown.

[0083] FIG. 26 shows an example heating port configuration in a packaged MEMS device having two front terminals (“Terminal 1” and “Terminal 2”). In the particular embodiment shown, the backside of the chip-scale package (CSP) is electrically coupled via a tethering structure to an electrode layer in a movable micromachined member (e.g., the resonant member shown at 119 in FIG. 1 ), thereby forming a third terminal, referred to herein as the “backside terminal” (“Terminal 3”). One of the front terminals is electrically coupled via another tethering structure to the same electrode layer as the backside terminal, such that the front and backside terminals, interconnected by the tether, form a heating port through which AC or DC Joule heating current is conducted. FIG. 27A shows an example circuit model of this Joule heating operation, where R tether The current I flows through the pair of tethers (modeled as JHThe current from terminal 1 to the lower (DDS) layer of the micromachined member also passes through one or more tether structures, which may be the same tether structures that conduct the Joule heating current. For all terminal connections, an equal tether resistance (R tether ), it should be noted that tether resistance may vary from layer to layer, even when a shared tether is used. For example, the cross-sectional area (width and / or height) of the tether and / or the electrical conductivity of the tether may be non-uniform in relation to its connection to different layers of the micro-machined member.

[0084] Still referring to FIG. 26 , the capacitor between terminal 1 and terminals 2 / 3 represents, in this particular example, a piezoelectric or other dielectric thin film within the movable micro-machined member. While a DC current source is coupled across terminals 2 and 3 in the illustrated configuration to generate Joule heating current (e.g., during fabrication or post-fabrication processing), a DC voltage source or an AC voltage or current source may alternatively be used. Also, after dicing a wafer (i.e., a wafer containing a plurality of such MEMS structures as shown in FIG. 26 ) into individual dies, the backside of each die may still be electrically coupled to one of the front terminals (i.e., through the tether and the resistance of the movable micro-machined member). In many applications, this residual electrical coupling (i.e., between the backside and front terminals) has no (or negligible) effect on device operation and may therefore be left intact. In applications requiring or benefiting from electrical isolation between the rear and front terminals of the device during device operation, a meltable or otherwise breakable connection to the rear terminal may be utilized to allow the rear terminal to be discarded after manufacture.

[0085] The two front terminals shown in FIG. 26 are utilized for normal device operation. In this example, the capacitors represent the piezoelectric layers, as in the case of the dual silicon piezoelectric MEMS resonator shown in FIG. 1A. During normal device operation, only one connection is required to each electrode layer (i.e., micromachined member) in the MEMS structure, and as a result, the backside of the MEMS die (terminal 3) may be unused. In the illustrated example model, the backside terminal is allowed to be electrically floating, so that no Joule heating current (or negligible Joule heating current) flows within the movable micromachined member. FIG. 27B illustrates this operation. In alternative embodiments, the backside terminal may be eliminated, as discussed above. Also, as with all embodiments herein, the micromachined member may be implemented / fabricated with more or less than the three layers shown in FIG. 26.

[0086] 28A and 28B show example electrical interconnections between encapsulation-level TSVs (e.g., through-silicon-vias extending through the encapsulation lid structure or substrate) and piezoelectrically actuated resonator electrodes that enable both package-level terminal interconnections of the MEMS resonator (i.e., in subsequent packaging steps) and Joule heating after encapsulation. In the embodiment of FIG. 28A, three terminals are exposed at the TSVs through the lid and / or substrate layer of the encapsulated dual silicon resonator: two terminals are bonded to opposite lateral ends of the top electrode (shown as ISDP layers, although any of alternative material layer configurations may be used), and one terminal is bonded to the bottom electrode (which in this example is a degenerately doped single crystal silicon layer). With this configuration, direct or alternating current may be passed through, for example, the resonator's tethering structure and top electrode layer, either or both of which may be designed to heat the resonator material stack (i.e., to Joule heat the resonator material stack described above) and thereby perform any number of post-encapsulation annealing and / or thermal frequency trim operations. The embodiment of Figure 28B is similar to that of Figure 28A, except that a Joule heating terminal pair is tethered to opposite lateral ends of the bottom electrode layer (e.g., a degenerately doped single-crystalline silicon layer) to operate the bottom tethering layer and / or bottom electrode layer as the heat source instead of the top layer tether / electrode element. In yet other embodiments, two Joule heating terminal pairs may be provided, each coupled to opposite ends of a respective degenerately doped silicon layer. Also, as described in further detail below, the joule heating terminal pair in any of the Figure 28A / Figure 28B embodiments may be merged with a third encapsulation terminal to form the resonator sense / drive terminal described with reference to Figure 1 A. Alternatively, the joule heating terminal pair may be driven by an equipotential resonator drive signal to emulate electrical coupling of the terminal pair.That is, instead of wiring the terminals together to establish an equipotential at the lateral ends of the associated material layers (as well as the tethering layers), the terminals may be driven to an equipotential by the drive / sense circuitry of the attached logic die.

[0087] 29A and 29B illustrate a terminal reduction scheme in which three or more encapsulation-level terminals (e.g., as described with reference to FIGS. 28A and 28B ) are merged prior to encapsulation within the package housing to expose only two electrically independent package-level terminals. In the conceptual diagram of FIG. 29A , the joule heating terminal (joule heating port) is coupled to the ISDP layer of the dual silicon resonator (which allows introduction of joule heating current through a wafer probe or singulated die probe), but this may alternatively be coupled to the core single crystal resonator layer or may be supplemented by an additional heating port coupled to the core resonator layer (i.e., in a four-terminal encapsulation structure).

[0088] 29B illustrates an exemplary sequence of packaging operations to produce the intra-package terminal merge (reduction) shown in FIG. 29A, showing cross-sectional and plan views of the encapsulation or packaging structure at each point in the sequence. Starting with an exemplary three-terminal encapsulation structure, two joule heating terminals are merged by metal deposition to form a unified top electrode terminal (or bottom electrode terminal). A further packaging layer (e.g., a polyimide or other non-conductive molding or material layer) is then disposed over the merged electrode terminals, with through vias and solder ball sockets formed as shown, to produce a four-terminal package in which two of the package-level terminals are electrically common (per terminal-merge metal deposition) to form the top electrode interconnection node, and a third of the four terminals is bonded to the bottom electrode (or vice versa). The fourth package-level terminal is provided primarily for mechanical stability and may be left unconnected ("no connect" or "NC") or may be redundantly coupled to the lower or upper resonator electrodes. In still other embodiments, particularly where a rectangular packaging configuration is desired, only two package-level contacts may be exposed—one for each of the two resonator electrodes.

[0089] In resonator-based systems for timing, sensing, or other applications, it may be desirable to calibrate the resonant frequency at or near room temperature to compensate for variability in the resonator manufacturing process. For example, the room temperature resonant frequency of an uncalibrated resonator may differ from its desired frequency by + / - 10,000 ppm (parts per million), compared to typical application requirements that are at the level of + / - 20 ppm or less.

[0090] The offset between the desired resonant frequency and the actual resonant frequency can be removed through active compensation (i.e., continuous correction) or through passive compensation (i.e., a one-time correction or "trim"). Active correction methods are undesirable in that they require dedicated electronic circuitry, consume power, and the compensated frequency output can be noisier than the uncompensated resonator frequency.

[0091] It is advantageous to perform frequency calibration once the resonator is hermetically packaged for several reasons. First, this can be performed in a non-cleanroom environment, thereby reducing costs. Second, the packaging process can shift the resonator frequency slightly, thereby reducing the accuracy of the frequency trimming process.

[0092] Passive compensation techniques have been developed to trim the room temperature frequency of MEMS resonators, for example through removal / addition of mass via laser irradiation or surface chemical reactions, but these have several limitations when applied to hermetically packaged resonators (e.g., damage to the package, generation of particulate matter within the sealed cavity, introduction of chemically reactive materials within the sealed cavity, increased cavity pressure, increased process complexity / cost, etc.).

[0093] In some embodiments disclosed herein, various Joule heating structures / techniques described above and / or alternative heating techniques (e.g., heating by laser or other electromagnetic radiation) are applied to achieve thermal frequency trimming operations. More generally, the application of heating methods involves altering one or more resonant frequencies or other resonant device characteristics based on altering material properties—alternative to the material removal, ablation, dimensional change, or surface modification described above. While these changes may occur as a side effect of heating to alter material properties, they are not the primary mechanism for achieving frequency trimming. One or more heating cycles in a sequence may be utilized to manipulate the resonant frequency of a MEMS structure. These and other thermal trimming embodiments and features are disclosed in further detail below.

[0094] FIG. 30 shows exemplary heating of a MEMS resonator via a laser (e.g., a continuous wave or pulsed laser with a wavelength ranging from ultraviolet to infrared) or other source of electromagnetic radiation. The MEMS resonator can optionally be hermetically encapsulated prior to frequency modification so that encapsulation-induced frequency shifts are compensated for. The hermetic enclosure can be fabricated from silicon, glass, alumina, or other materials with minimal optical absorption at a particular wavelength compared to the DDS housing the MEMS structure. For example, an IR laser (e.g., at a 1.1 micron wavelength) can be heated in a lightly doped silicon hermetic enclosure (e.g., 1e17 atoms / cm) such that the absorption of the DDS layer in the MEMS resonator (e.g., implemented by the dual silicon resonator or variations described above) is much greater than in a lightly doped silicon enclosure. 3) can be combined with a laser. The laser can heat the resonator while leaving the hermetic encapsulation relatively intact. In the case of a laser beam, the spot size can be substantially round or can have alternative shapes through optical manipulation. The spot size will typically be much smaller than the MEMS structure, but in some cases it can be beneficial to use a laser spot comparable in size to the structure. The laser spot can be applied to one or more discrete locations on the structure, or it can be scanned to heat a specific area, shape, or pattern.

[0095] In the case of heating via electromagnetic radiation, the DDS layer can perform two functions: frequency tuning through stiffness modification and optical absorption. Alternatively, the DDS layer can be used for only a single function (optical absorption), and the frequency of the MEMS structure can be altered through laser ablation, resulting in a shape change that reduces stiffness or mass depending on the spatial distribution of the ablation. Optical power can be spatially patterned across the MEMS structure, for example, to independently engineer the room-temperature frequencies of two or more modes or to independently control the frequency temperature coefficients of two or more modes.

[0096] In laser-based heating methods, the tether performs two functions (thermal / thermal insulation and mechanical compliance), and its potential third function of providing a relatively large electrical resistance is of less importance. The heating power source is located at the point of laser incidence, not within the hot spot of the tether or patterned electrodes, as in the case of Joule heating.

[0097] 31A and 31B illustrate the manipulation of electrically active dopant concentration through heating and cooling processes applied to MEMS resonator structures. The total dopant concentration in a semiconductor is the sum of the electrically active and non-electrically active dopant concentrations. In DDS materials, the electrically active dopant concentration at the completion of wafer-scale processing can be significantly less than the total dopant concentration through multiple dopant deactivation mechanisms.

[0098] The DDS fabrication process can be engineered so that at the completion of the wafer fabrication process, a high concentration of dopants is electrically inactive. The inactive dopant concentration can be controlled through the choice of dopant species (e.g., arsenic, boron, phosphorus, or antimony), dopant dose / unit area, dopant concentration distribution, or thermal processing parameters (e.g., time, temperature, heating rate, cooling rate) applied to the resonator.

[0099] Figure 16A shows the electrically active dopant concentration after such a DDS sample (e.g., degenerately doped single crystal silicon or degenerately doped polycrystalline silicon) is heated. The electrically active dopant concentration does not change significantly at low temperatures (e.g., below 200°C), but increases significantly at relatively high temperatures, up to the point where a large proportion, or potentially substantially all, of the previously inactive dopant becomes thermally activated. The maximum heating temperature can range anywhere from 600°C to 1300°C.

[0100] Figure 31B shows two potential scenarios for the electrically active dopant concentration as the DDS cools. For a sufficiently slow cooling rate (e.g., 1 °C / min), the electrically active dopant concentration will decrease as the dopant deactivates through the formation of inactive clusters, for example, in the case of phosphorus doping. After slow cooling to room temperature, a large proportion of the dopant atoms may become inactive. In contrast, if the DDS is cooled at a rate of 10 4Rapid cooling ("quenching") at rates on the order of °C / second (smaller or larger cooling rates may be effective) results in deactivation of a relatively small number of dopant atoms. Subsequent heating to a relatively lower temperature or a relatively slow cooling rate may be used to decrease the electrically active dopant concentration, thereby increasing or decreasing the dopant concentration. Thus, cycles of heating followed by rapid cooling ("heating cycles" or "heating cycles" or "heating / quenching cycles") can be used to manipulate the electrically active dopant concentration in a DDS.

[0101] Figure 32A shows the relationship between resonant frequency and the electrically active dopant concentration of a DDS. The electrically active concentration of carriers in a DDS is related to its stiffness (e.g., c in silicon) at room temperature. 11 , c 12 and c 14 Both the temperature coefficient (temperature coefficient) and its first, second, and higher order stiffness temperature coefficients can be varied. Thus, varying the electrically active dopant concentration of a DDS can be utilized for frequency trimming.

[0102] Figure 32B shows the variation of room temperature resonant frequency with heating time and heating temperature. The rate of change of frequency varies exponentially with heating time and typically saturates to a steady-state frequency for sufficiently long heating times. An increase in heating temperature is typically associated with a decrease in resonant frequency. However, as mentioned above, after heating a MEMS device to a particular temperature, it is possible to potentially decrease the active dopant concentration (increase the resonant frequency) by subsequently heating the device to a lower temperature, a technique referred to herein as "back-tuning."

[0103] Temperature-activated material modification of MEMS structures is not specific to a particular transduction technique. For example, it can be easily applied to electrostatically, piezoelectrically, piezoresistively, or magnetically transduced devices. Likewise, it is not specific to Joule heating or laser-based heating. Any other heating method capable of controllably heating MEMS structures to sufficiently high temperatures may be applied. In the case of laser beam heating, it may be advantageous in some cases to position the laser beam in areas not covered by electrodes if electrodes are placed on the resonator. For example, piezoelectric resonators can use electrodes on the resonator. The resonator may be engineered to position certain areas of interest, such as areas experiencing large strains or large movements in vibrational modes, away from the electrodes to avoid damage to the electrode surface associated with laser illumination.

[0104] 33 shows an exemplary sequence of heating (power) pulses applied sequentially to a MEMS resonator to repeatedly change its resonant frequency. In this particular example, after three similar heating pulses are applied to the structure, the frequency has saturated and is no longer decreasing. The heating power and duration of the fourth pulse are increased, and a further decrease in the resonant frequency is achieved.

[0105] The temperature of a MEMS structure in a thermal process can be controlled by varying the applied power and / or the duration of the applied power level. For example, applying a heat pulse having a duration of 1 nanosecond (nsec) to a MEMS structure with a characteristic thermal time constant of 1 millisecond (msec) will result in a much lower maximum temperature than a 100 millisecond pulse, because in the former case the structure will never reach thermal equilibrium. The thermal time constant of a MEMS structure can be engineered through the geometry or materials from which the structure is constructed.

[0106] FIG. 34 is a flowchart illustrating an example frequency trimming procedure or algorithm. The routine begins by measuring the resonant frequency of the resonator at one or more temperatures. The temperature can be adjusted using an external source or by passing a current through the device (i.e., any of the radiative or Joule heating techniques disclosed above, and more generally, any feasible technique for heating a MEMS structure). If the resonator is not operating in a desired manner, the heating power, duration of exposure, and spatial distribution of the thermal process are calculated to alter the resonator characteristics. The thermal process is applied to the MEMS structure, for example, through Joule heating or laser radiation. After waiting a period of time until the MEMS structure reaches a desired temperature, such as the temperature of the surrounding room, the resonant frequency is again measured to determine whether the desired operation has been achieved. The resonant frequency may be measured when the resonator is in a hot state, but there will consequently be an offset between the measured frequency and the room temperature frequency. Therefore, it is preferable to measure the resonant frequency at a temperature representative of the target application.

[0107] In practice, frequency trimming consists of multiple repeated cycles (iterations) of measuring the resonant frequencies of one or more eigenmodes at one or more temperatures and then applying a thermal process to the MEMS structure to change the room-temperature frequency or TCF of one or more modes. By varying the thermal process characteristics (e.g., maximum temperature or duration) or the spatial distribution of the thermal process (e.g., laser position, AC frequency, or effective electrical terminals), multiple parameters (e.g., the frequencies of two different resonant modes, or the room-temperature frequency and first-order TCF) can be independently varied. During the thermal measurement cycles that comprise each iteration, it is beneficial, but not necessary, to wait a sufficiently long time after the heating cycle for the MEMS structure to cool to ambient or a known temperature before measuring the frequency. The temperature of the MEMS structure can be estimated by passing a relatively small current through the structure to measure its electrical resistance, which changes with temperature. Alternatively, various optical or other non-contact temperature measurement techniques (e.g., vibrometers to identify resonant frequencies following application of power pulses) may be used to determine the overheat temperature, and / or in-situ temperature sensing structures may be used to determine the overheat temperature, but once detected, the overheat temperature may be fed back to a heating control circuit, such as to enable closed-loop heating of the MEMS structure to a target frequency and to control the heating and / or quenching rate / profile.

[0108] After each thermal measurement cycle, parameters for the next cycle can be calculated. This process may be aided by a model-based approach, where the expected frequency change for the thermal process is calculated and compared to the measured frequency change. If the magnitude of the frequency change in a particular iteration is smaller than expected, the thermal power of the next iteration can be increased to compensate. Similarly, a larger-than-expected frequency change can be used to modify the thermal power of the next iteration. This feedback process can reduce the number of iterations and, therefore, the cost of frequency trimming.

[0109] FIG. 35 shows an example frequency trimming procedure. At the start of each iteration, the resonant frequency is measured and the thermal process to be applied is calculated. Initially, the frequency is above the target frequency. Initially, the thermal power is small to avoid over-adjusting the frequency and potentially calibrate the mathematical model to the characteristics of this particular resonator. Over a total of seven iterations, the thermal power is gradually increased and the frequency is gradually decreased until the target frequency is achieved. The frequency trimming process may be terminated when the resonant frequency falls within an acceptable range, or alternatively, execution can be stopped if the number of iterations exceeds a prescribed limit or if the frequency becomes too far from the target frequency for a high probability of success.

[0110] FIG. 36 shows an example frequency trimming procedure that includes a frequency back-tuning technique. As mentioned above, the thermal frequency tuning of a DDS can be reversible in certain circumstances. The ability to reverse the tuning process allows for the use of a more aggressive frequency trimming algorithm. In this example, the thermal power is rapidly increased. After the third cycle, the resonant frequency is below the target frequency. The remaining two iterative cycles reduce the thermal power to increase the frequency until the target frequency is achieved.

[0111] To save time and reduce manufacturing costs, multiple structures can be trimmed simultaneously. For example, an electrical probe station can be constructed to measure and trim two, four, eight, or various other numbers of devices simultaneously. Alternatively, or in combination, multiple structures can be electrically contacted simultaneously and trimmed sequentially.

[0112] The frequency tuning process may be used to simultaneously trim multiple system parameters and is not limited to the resonant frequency of a single resonant mode at room temperature. For example, the room temperature of multiple resonant modes, the temperature coefficients of a single mode, or any combination thereof can be adjusted. This is possible by controlling the spatial distribution of temperature across the MEMS structure in a thermal modification process. For example, consider two locations on a resonator, P1 and P2, and two vibration modes, M1 and M2. The change in frequency of an individual eigenmode due to a change in stiffness at a particular location is proportional to the mechanical stress at that location. For example, the mechanical stress at P1 may be small for M1 and large for M2. The opposite may be true for P2 relative to M1 and M2. Therefore, changing the stiffness at P1 will preferentially change the frequency of M1, and changing the stiffness at P2 will preferentially change the frequency of M2. This concept of discrete locations can be generalized to continuous temperature and stress distribution across a MEMS structure.

[0113] Exemplary methods of controlling and altering the temperature distribution across a MEMS structure include varying the position of the incident laser beam, changing the selection of active and reactive electrical terminals used for heating, or changing the frequency of the AC input to change the distribution of current flowing through the device, thereby resulting in an increase in temperature in relation to the current passing through the distributed parasitic capacitance.

[0114] FIGS. 37A-37D illustrate exemplary manipulation of multiple resonant frequencies in a MEMS structure based on heating by a subset of all available heating terminals. FIG. 37A shows a simplified disk-ring gyroscope with two vibration modes of interest, Mode 1 and Mode 2. FIG. 37 illustrates an exemplary array of electrodes surrounding a MEMS structure, potentially utilized in capacitively coupled Joule heating as well as for use in normal device operation. FIG. 37C illustrates RF input applied to a subset of electrodes to non-uniformly heat the MEMS structure to preferentially tune the frequency of Mode 1, and FIG. 37D illustrates RF input applied to a different subset of electrodes to preferentially adjust the frequency of Mode 2 to achieve a desired frequency relationship between the two modes. Temperature-activated frequency trim can be used to manipulate the sum or product of two or more vibration modes. Example applications include, without limitation, mode matching, engineering the frequency difference between two modes, and controlling the frequency ratio of the two modes.

[0115] Figure 38 shows the tuning of the frequency relationship between two modes of a similar MEMS structure based on activation of a subset of heating terminals, except that Joule heating is achieved through terminals directly coupled to the structure, thereby allowing the possibility of DC heating current.

[0116] The DDS layer or the dopant concentration within the DDS layer can be varied across a MEMS structure. For example, regions of high electrical resistivity can be introduced within the structure to increase local Joule heating power density. In another application, varying the dopant concentration across the structure can allow for localized changes in heat-induced mechanical stiffness, even when the temperature is relatively uniform across the structure. This can be used to engineer one or more temperature coefficients of frequency (TCF) of the structure. For example, a MEMS resonator can be formed from two or more regions that are mechanically and thermally coupled but have substantially different TCFs. The room-temperature frequency and TCF of the composite structure will depend on the weighted contributions from each region. Therefore, changing the mechanical stiffness of one region relative to another will result in a change in the TCF of the composite structure.

[0117] Asymmetrical changes in the stiffness of MEMS structures can also be used to induce changes in the quality factor (Q) of resonator devices. Asymmetrical changes in the mass or stiffness of a resonator result in increased energy loss at the point where the resonator connects to the substrate or package, resulting in a reduced Q. If the resonator is designed with asymmetry and the laser modifies the resonator to remove the asymmetry, the Q can be increased. The Q level of the system can be calibrated by adjusting the spatial distribution of temperature during thermal treatment, for example, through the laser position or through the active electrical terminals.

[0118] The various circuits and MEMS structures disclosed herein may be described in terms of their operation, register transfers, logic components, transistors, layout geometries, and / or other characteristics using computer-aided design tools and expressed (or otherwise represented) as data and / or instructions embodied in various computer-readable media. File and other object formats in which such circuit representations may be implemented include, without limitation, formats supporting behavioral languages such as C, Verilog, and VHDL; formats supporting register-level description languages such as RTL; and formats supporting shape description languages such as GDSII, GDSIII, GDSIV, CIF, MEBES, and any other suitable formats and languages. Computer-readable media on which such formatted data and / or instructions may be embodied include, without limitation, various forms of computer storage media (e.g., optical, magnetic, or semiconductor storage media, whether independently distributed by that manner or stored "in situ" within an operating system).

[0119] When received within a computer system via one or more computer-readable media, such data and / or instruction-based representations of the circuitry described above may be processed by a processing entity (e.g., one or more processors) within the computer system in conjunction with the execution of one or more other computer programs, including, without limitation, netlist generators, place and route programs, and the like, to generate a representation or image of the physical manifestation of such circuitry. Such representations or images may then be used in device fabrication, for example, by enabling the generation of one or more masks used to form various components of the circuitry in the device fabrication process.

[0120] In the above description and in the accompanying drawings, specific terminology and drawing symbols are set forth to provide a thorough understanding of the disclosed embodiments. In some instances, the terminology and symbols may imply specific details that are not required to practice the embodiments. For example, specific dimensions, quantities, temperatures, periods, signal levels, power levels, signaling or operating frequencies, component circuits or devices, and the like, may differ from those described above in alternative embodiments. In addition, links or other interconnections between integrated circuit devices or internal circuit elements or blocks may be shown as buses or as single signal lines. Each bus may alternatively be a single signal line, and each single signal line may alternatively be a bus. However, the signals and signaling links shown and described may be single-ended or differential. A signal driver circuit is said to “output” a signal to a signal receiver circuit when the signal driver circuit asserts (or, if the context so explicitly indicates or indicates, deasserts) a signal on a signal line coupled between the signal driver circuit and the signal receiver circuit. The term “coupled” is used herein to describe not only a direct connection, but also a connection through one or more intervening circuits or structures. Device “programming” can include, for example, and without limitation, loading control values into registers or other storage circuits within an integrated circuit device in response to a host instruction (and thus controlling device operational aspects and / or establishing device configuration) or through a one-time programming operation (e.g., blowing fuses in configuration circuitry during device manufacture) and / or connecting one or more selected pins or other contact structures of the device to a reference voltage line (also referred to as strapping) to establish a particular device configuration or device operational aspect. The terms "exemplary" and "embodiment" are used to express an example, not a preference or requirement.Also, the terms "may" and "can" are used interchangeably to denote optional (permissible) subject matter. The absence of either term should not be construed to mean that a given feature or technique is required.

[0121] Various modifications and variations may be made to the embodiments presented herein without departing from the broad spirit and scope of the present disclosure. For example, a feature or aspect of any of the embodiments may be applied in combination with any other of the embodiments, or in place of a comparable feature or aspect thereof. Accordingly, the specification and accompanying drawings are to be regarded in an illustrative and not a restrictive sense.

Claims

1. 1. A microelectromechanical systems (MEMS) device comprising: A substrate; a movable micro-machined member; a mechanical structure securing the movable micro-machined member to the substrate, insulating the movable micro-machined member from the substrate, and providing a conductive path to allow heating of the movable micro-machined member to a temperature of at least 300°C; An apparatus comprising:

2. The MEMS device of claim 1 , further comprising an encapsulation structure forming a vacuum cavity in which the movable micro-machined member and at least a portion of the mechanical structure are disposed.

3. The MEMS device of claim 1 , wherein the mechanical structure comprises a folded tether.

4. The MEMS device of claim 1 , wherein the movable micro-machined member comprises a MEMS resonator.

5. 10. The MEMS device of claim 1, wherein the mechanical structure securing the movable micro-machined member to the substrate provides sufficient thermal insulation and electrical resistivity to enable heating of the movable micro-machined member to a temperature of at least 700°C when a Joule heating current is conducted through the conductive path.

6. 10. The MEMS device of claim 1, wherein the mechanical structure includes an electrode coupled in series with and forming a portion of the conductive path, the electrode being patterned to include a resistive region having sufficient resistance to heat the movable micro-machined member to a temperature of at least 300°C when a current is conducted through the conductive path.

7. The MEMS device of claim 6 , wherein the electrodes comprise degenerately doped polycrystalline silicon.

8. The MEMS device of claim 6 , wherein the electrodes comprise degenerately doped single crystal silicon.

9. 7. The MEMS device of claim 6, wherein the resistive region includes a via feature of the electrode extending between and electrically coupling first and second terminal regions of the electrode, the via feature having a cross-sectional area significantly smaller than either of the first or second terminal regions and coupled in series with the first and second terminal regions in the conductive path.

10. The MEMS device of claim 1 , wherein the movable micro-machined member comprises a degenerately doped semiconductor.

11. 1. A microelectromechanical system (MEMS) resonator comprising: a layer of degenerately doped polycrystalline silicon; a layer of degenerately doped single crystal silicon; a layer of piezoelectric material sandwiched between the degenerately doped polycrystalline silicon layer and the degenerately doped single crystalline silicon layer; A resonator comprising:

12. The MEMS resonator of claim 11 , wherein the layer of piezoelectric material comprises aluminum nitride.

13. 12. The MEMS resonator of claim 11, wherein the layer of degenerately doped single crystal silicon is at least ten times thicker than the layer of piezoelectric material.

14. 12. The MEMS resonator of claim 11 , wherein the layers of degenerately doped polycrystalline silicon, degenerately doped single crystal silicon, and piezoelectric material form a resonator body and one or more tethering structures that mechanically couple the resonator body to anchoring points within a field area.

15. 15. The MEMS resonator of claim 14, wherein the layer of degenerately doped polycrystalline silicon includes a first electrode within the resonator body and a first conductive path within each of the one or more tethering structures to enable the first electrode to be electrically coupled to a first node of a voltage supply external to the resonator through at least one of the tethering structures.

16. 16. The MEMS resonator of claim 15, wherein the layer of degenerately doped single crystal silicon includes a second electrode within the resonator body and a second conductive path within each of the one or more tethering structures to enable the second electrode to be electrically coupled to a second node of the voltage source external to the resonator such that a voltage generated by the voltage source provides an electrostatic potential across the layer of piezoelectric material.

17. 16. The MEMS resonator of claim 15, wherein the one or more tethering structures include at least two tethering structures that enable the first electrode to be coupled between the first node of the voltage supply source external to the resonator and a second node of the voltage supply source external to the resonator such that a potential difference is applied across the first electrode.

18. 16. The MEMS resonator of claim 15, wherein the first electrode is patterned to form a resistive region having sufficient resistance to heat the resonator body to a temperature of at least 300°C when a current is conducted through the resistive region.

19. 16. The MEMS resonator of claim 15, wherein the one or more tethering structures have sufficient electrical resistivity to insulate the resonator body from the anchoring points and to enable heating of the resonator body to a temperature of at least 300° C. when a Joule heating current is conducted through the one or more tethering structures.

20. The MEMS resonator of claim 15 , wherein the one or more tethering structures include two tethering structures coupled to the first electrode on opposite sides of the resonator body.

21. 1. A method for adjusting a resonant frequency of a microelectromechanical system (MEMS) resonator, comprising: heating the MEMS resonator to a temperature of at least 700° C. in a heating operation; cooling the MEMS resonator to a temperature less than 100° C. in a quenching operation after the heating operation; determining a frequency of at least one resonant mode of the MEMS resonator in a frequency measurement operation after the quenching operation; repeating the heating, quenching, and frequency measurement operations if a difference between the frequency determined in the frequency measurement operation and a target frequency is greater than a threshold; A method comprising:

22. 22. The method of claim 21, wherein heating the MEMS resonator in the heating operation comprises joule heating the MEMS resonator after the MEMS resonator is hermetically encapsulated.

23. 22. The method of claim 21 , wherein heating the MEMS resonator in the heating operation comprises heating the MEMS resonator with a radiative heat source after the MEMS resonator is hermetically encapsulated, wherein radiation from the radiative heat source passes through structural elements used to encapsulate the MEMS resonator.

24. 22. The method of claim 21 , wherein heating the MEMS resonator to a temperature of at least 700° C. in the heating operation comprises dissipating energy within one or more features of the MEMS resonator, the one or more features comprising at least one of a tethering structure mechanically coupling to a field area surrounding a body of the MEMS resonator and a patterned electrode disposed on or within the body of the MEMS resonator.

25. 22. The method of claim 21, wherein heating the MEMS resonator to a temperature of at least 700°C in the heating operation comprises applying increasing amounts of power to the MEMS resonator to control a rate at which the MEMS resonator heats.

26. 22. The method of claim 21 , further comprising: determining a temperature to which the MEMS resonator is heated in the heating operation; and adjusting an amount of power applied in the heating operation until the MEMS resonator temperature reaches a target temperature.

27. 22. The method of claim 21, wherein cooling the MEMS resonator to a temperature less than 100 degrees Celsius in the quenching operation comprises cooling the MEMS resonator to a nominal room temperature.

28. 22. The method of claim 21, wherein cooling the MEMS resonator in the quenching operation comprises controlling a rate at which the MEMS resonator cools.

29. 22. The method of claim 21, wherein if the frequency difference is above the threshold, repeating the heating, quenching, and frequency measurement operations comprises repeatedly repeating the heating, quenching, and frequency measurement operations until the frequency difference no longer exceeds the threshold.

30. 30. The method of claim 29, wherein in at least one iteration of the heating, quenching, and frequency measurement operations, one or more of a rate at which the MEMS resonator is heated, a temperature to which the MEMS resonator is heated, a rate at which the MEMS resonator is cooled, or a temperature to which the MEMS resonator is cooled is different from another execution of the heating, quenching, and frequency measurement operations.

Images