Three-dimensional (3D) direct writing based on the meniscus-confined electrodeposition of copper metal wires was used in this study to develop vertical capacitive microelectromechanical switches. Vertical microelectromechanical switches reduce the form factor and increase the area density of such devices in integrated circuits. We studied the electromechanical characteristics of such vertical switches by exploring the dependence of switching voltage on various device structures, particularly with regard to the length, wire diameter, and the distance between the two wires. A simple model was found to match the experimental measurements made in this study. We found that the electrodeposited copper microwires exhibit a good elastic modulus close to that of bulk copper. By optimizing the 3D structure of the electrodes, a volatile electromechanical switch with a sub-5 V switching voltage was demonstrated in a vertical microscale switch with a gap distance as small as 100 nm created with a pair of copper wires with diameters of ~1 μm and heights of 25 μm. This study establishes an innovative approach to construct microelectromechanical systems with arbitrary 3D microwire structures for various applications, including the demonstrated volatile and nonvolatile microswitches.
Microelectromechanical systems (MEMS) have been used in many applications in many industries, including the medical, automotive, optics, electronics, and biotechnology sectors1,2. MEMS switches operated by radio frequencies are essential components in the circuit architectures of a variety of MEMS-based systems3. As an electromechanical switch, MEMS switches offer markedly lower power consumptions, better isolations, and lower insertion losses compared with conventional field-effect transistors or PIN-diode-based switches3,4. MEMS switches are typically fabricated with silicon-surface micromachining technology and are typically planar, which requires a large surface budget on integrated circuit chips5. Submicron floating cantilevers fabricated with a metal layer (for example, Au, Al, Pt, and so on) are then used as mechanical moving electrodes. However, in thin metal films, it is difficult to control the stress that causes most metal cantilevers to warp upon release6,7. To accommodate the further miniaturization and integration of devices and systems, there is a trend towards developing three-dimensional (3D) circuit architectures and related 3D devices, including novel 3D MEMS switches.
Vertical switches that use aligned nanowires and nanotubes have exhibited outstanding capacities in MEMS switches. The simplest microwire-based electromechanical switch consists of two parallel and freestanding microwires that are electrically connected to fixed electrodes. When a voltage is applied between the two microwires, the induced electrostatic force pulls the wires towards each other. The threshold switching voltage is thus defined as that voltage that establishes an ‘on state’ when a higher voltage forms electrical contact between the deflected wires. These wires exhibit high stiffness and large deflection capacities to achieve ultrafast switching response times8. In addition, the micro/nano-scale size of these wires is favorable to achieve ultrahigh density and ultralow-energy-consumption switch systems9. As a result, capacitive switches that are based on nanowires and nanotubes have been popular for use as robust nanotweezers via electrostatic defection10,11. For example, researchers have integrated silicon nanowires12,
Metal contact switches have a low on-resistance and a high isolation and are suitable for signals with frequencies including DC, millimeter wave (that is, 100 GHz) and others32. It is expected that reducing the gap distance between these wires using 3D structures in switching devices is the most effective means to reduce the threshold switching voltage and thus resolve electrical burnout issues. Among available metals, success in controlling the size and shape of the wires produced has primarily limited the materials used to gold and silver, which has led to a wide range of applications33,
In this study, we propose the use of a high-efficiency, direct-writing technology that is based on the meniscus-confined electrodeposition principle37 to fabricate vertical microelectromechanical switches. We also report on the electrostatic pull-in behavior of 3D-microswitch-based on Cu MWs for the first time. Meniscus-confined electrodeposition has been demonstrated to be an effective direct-writing process for fabricating intricate 3D metal microstructures that do not use complex lithography processes38. The vertical Cu MWs were electrodeposited onto surface electrodes using the meniscus-confined electrodeposition method. A glass substrate lithographically patterned with Au/Ni surface electrodes was used as the base for growing vertical Cu MWs. The height and shape of the Cu wires are defined by the designed motion path of the translation stages, and the diameter of the wire is defined by the nozzle size of the glass micropipette used in the system. Cu is used due to its good electrical and mechanical properties.
Materials and methods
Figure 1 shows the proposed 3D design and the fabrication steps required to produce this design. The fabricated device consists of two flexural copper microwires (Cu MWs) that are perpendicular to the substrate surface on the surface electrodes. The microwires were constructed on conducting (gold) surfaces via a meniscus-confined 3D electrodeposition process (that is, a 3D direct-writing process), resulting in the fabrication of complex metallic microarchitectures. It is expected that reducing the gap distance d of the vertical aligned microwire to the submicron/nano-scale is required to lower the threshold switching voltage and thus resolve electrical burnout issues in switching devices. This process is challenging and less consistent for vertically aligned 1D microwires. Using intricate 3D wire structures, the proposed design makes the electrodepositional growth of Cu MWs on the substrate more straightforward and convenient to reduce d between the microwires. The apparatus shown in Figure 1a consists of an electrolyte-containing micropipette with a microscopic dispensing nozzle, a high-resolution monitoring system, electrical equipment and a central control unit. The glass micropipette, which has an opening diameter of 1 μm, is produced by a pipette puller (P-2000, Sutter Instrument Co., Novato, CA, USA) and is filled with a Cu electrolyte (0.1 M CuSO4). The conductive substrate surface (that is, anode) is created at the rear opening of the pipette, and gold-coated glass wafers are used as substrates. The direct-writing process is conducted at a constant relative humidity of 80%. Both the positioning and the pulling speed of the pipette are accurately controlled with a three-axes high-precision nanopositioning piezo stage (P-562.3CD, Physik Instrumente GmbH, Karlsruhe, Germany) with a positioning resolution below 10 nm. An electrical potential between the Au electrode and the substrate was applied using a high-voltage-source measurement unit (237, Keithley, Cleveland, OH, USA). The printing process was observed in situ using a high-resolution monitoring system that consisted of an optical objective lens (VMU-V, Mitutoyo, Kawasaki, Japan) and a 3.3-megapixel digital camera (SC30, Olympus, Tokyo, Japan). As shown in Figure 1c, switching devices with different vertical segment lengths L1, slant segment lengths L2, slant angles α and distances D were fabricated. In this study, Cu MWs with defined diameters 2r of ~1.0 μm were fabricated and used as active components in the device. The detailed description of the metal microwires growth by this technique can be found elsewhere38,39.
The microscopic characteristics of the printed microswitch architectures were analyzed using field emission scanning electron microscopy equipped with an energy dispersive spectrometer (SEM-EDS, FEI QUANTA 250 FEG). The pull-in/pull-out characteristics of the device were investigated by an electromechanical analysis. As shown in Figure 1b, the bottom electrodes are electrically connected to a power supply. A resistor was connected in series with the device to limit the current at the ‘on state’ passing through the device25. As the voltage applied across the active element, the resulting electrostatic force is balanced by elastic restoring forces from the deflection of the Cu MWs. When a critical ‘pull-in’ voltage is reached, the electrostatic forces overwhelm the remaining force, and the microwires make electrical contact with each other, establishing the ‘on state’. The switching voltage V will be determined by the balance of the electrostatic force between the wires and the elastic force of the wires. The switching voltage was measured by monitoring the sudden rise in the current passing through the device. All electrical measurements were performed at room temperature and pressure on a probe station with a flow of dry nitrogen flooding the inner chamber. The size dependencies of the pull-in voltage were predicted by the finite element method.
Results and discussion
Manufacturing 3D Cu microswitches on device substrates
To demonstrate the proposed technique, the effectiveness of the meniscus-confined electrodeposition for Cu-based microswitches were studied and evaluated. In the proposed process, Cu MWs with diameters of ~1 μm and heights of 20 μm are grown using a micropipette with a diameter of ~1 μm filled with 0.1 M CuSO4 aqueous solution. Because metal growth is confined within the meniscus, which can be between several microns and 100 nm in size, the number of nucleation sites and the size of the grains are controlled by small-scale electroactive areas during growth. We have confirmed that the nanowire has a nanocrystalline structure with an average grain size of tens nanometres39. The nanocrystalline structure results in a smooth surface37,39, and ultrahigh elastic strains and fracture strengths were reported to be sustained in Cu nanowires due to grain boundary strengthening40,41. This feature can be beneficial for the operating reliability of MEMS switches. Three-dimensional (3D) metallic microarchitectures with diverse and complex features can be printed by accurate handling of the electrolyte meniscus during the printing process. As shown in Figure 2, complex 3D electrodes with high aspect ratios and truly 3D microstructures were built at desired locations by moving the electrode with respect to the substrate. The different lengths and angles used depended on the different moving path applied, which can be realized using an automated computer-controlled system. The electrodes were found to stick to each other as the electrode distance decreased to ~0.5 μm. As shown in Figure 2b, steric constraints between the glass micropipette and the adjacent electrode will occur as the active gap d decreases due to the shape of the glass micropipette used. When d decreased to the threshold distance, the liquid bridge between the dispensing nozzle and the adjacent electrode becomes more prevalent due to capillary forces, causing irreversible adhesion to occur (for example, the rightmost pair Cu MWs shown in Figure 1a). Thus, the gap distance should satisfy d>L1 tan θ, where θ is the slant angle of the pipette wall shown in Figure 2b. In the experiments in this study, θ is ~3° for the 1-μm-diameter micropipette, and the gap distance should be more than approximately 520 nm to avoid adhesion. It has been reported that reducing the gap distance can effectively decrease the pull-in voltage and mitigate damage in switching devices9,17,42,43. As a result, it is impossible to further decrease the active distance. As discussed in the following section, we can decrease the switch-on voltage to several volts by optimizing the 3D electrode structures.
The propensity of Cu to oxidize under ambient conditions is an imminent challenge for the applicability of Cu nanostructures. To prevent surface oxidation, the synthesis of Cu nanostructures must be performed under inert atmosphere and reducing environments. However, the energy dispersive spectrometer (EDS) maps of copper and oxygen elements in a single microwire, as shown in Figure 2c, provide evidence for the marked oxidation resistance of the proposed process, despite fabrication occurring in ambient conditions. Thus, Cu MWs were produced via the 3D direct-writing process. In general, devices at this scale are sufficient for reliable operation and special needs; structural optimization is used, as shown in the following section.
Nonvolatile switching behavior
Figure 3a shows the SEM image of a typical switch device operated at a high switching voltage due to the initially large gap distance. The I−V curve shown in Figure 3c shows a nonvolatile switching behavior with a threshold voltage of 46.5 V, which is higher than other reported Ge-based devices21 and carbon-nanotube-based electromechanical switches20. Upon switching to the ‘on state’, an Ohmic contact between the two Cu wires was observed with an overall contact resistance measured to be ~2.0 MΩ. The microwires did not return to their original position and remained stuck together, even after the bias voltage was removed. It is known that a high bias voltage can induce damage to the switching device9,26. The high-magnification SEM image shown in Figure 3b indeed shows two joining bridges in the contact area between the wires, indicating a burnout, which resulted in effectively spot welding the two wires together. The van der Waals attractive force between the wires was found to not be responsible for the bonding of the two wires both from a simple quantitative estimation and experimentally. In the experiment, one of the Cu wires was pushed into contact with the other wire using a manipulator; no clear bonding was observed. This feature of maintaining a permanent electrical connection with the application of a set bias could be useful in microdevices that require nonvolatile switching.
Tuning the design parameters of the nonvolatile switches
To optimize the design parameter of the switches, we studied the electromechanical response of the Cu microwire switch experimentally and analytically. To investigate the instability of the switches, the numerical finite element method was used. The total potential energy of the system consists of the contributions from the mechanical energy44,
where E is the Young’s modulus of Cu, r is the radius of the wire, x is the vertical coordinate of the microwire, u is the deflection of the microwire from the original position, V is the applied static voltage between the microwires, C is the capacitance between the two Cu MWs, which is defined as follows47,48:
where εr and ε0 are the relative permittivity of air (εr=1) and the permittivity of vacuum (ε0=8.854×10−12 F m−1), respectively. Using the minimum energy principle, δE(u)=0, the following equation can be obtained:
The dimensionless form of the governing equation of Cu MWs is shown above, and the boundary value problem for each microwire can be defined as follows:
Figure 4 shows the modeled result and its comparison with experimental measurements. These devices were fabricated, and their dimensions were calibrated in a scanning electron microscope. Their switching behaviors were then measured in a clean ambient environment. Figures 4a and b depict the dependence of the pull-in voltage of the device on the segment length of the microwires. As expected, the required pull-in voltage decreases as the segment lengths increase due to the reduced elastic stiffness of the microwires. The pull-in voltage is shown to be more sensitive to changes in the segment length L1 due to the narrower gap distance between them, thus creating a larger electrostatic force. Figure 4c shows the effects of the angle α on the pull-in voltages of the device. As the angle α increases, the horizontal bending stiffness of the L2 segment decreases in the u direction, and consequently, the required pull-in voltage slowly decreases. The primary reason for this phenomenon is that the moment of inertia decreased due to a change in the horizontal cross-section of the vertical wires49. The pull-in voltage is shown to be most sensitive to the distance between the L1 segments, as shown in Figure 4d, because the capacitance between the microwires exhibits a virtually inverse dependence on the gap distance and thus the electrostatic force upon the applied voltage. Electrostatically actuated test structures can describe the MEMS system’s material property43. In all cases, the modeling results agreed with the experimental results, indicating both the good mechanical and electrical characteristics of the fabricated microstructure.
To describe the effect of the Cu MWs’ diameters on the pull-in voltage in detail, a total of 22 devices with different diameters were investigated; the detailed pull-in voltages of those devices are shown in Figure 5. In this study, the diameters of Cu MWs varied from ~0.4 to 1.6 μm, and the response of the pull-in voltage was approximately 10 V per 100 nm. It was found that the pull-in voltages of the devices increase as their diameters increase. This change in the pull-in voltage with the diameter is relatively slow when the diameters of the Cu MWs are small. From Equation (3), the pull-in voltage approximately follows as the gap distance d approaches the diameter 2r. Concurrently, the gap distance exhibits an approximately linear change in the pull-in voltage when the diameter is constant, which is also shown in Figure 5. This result agrees with the description shown in Figure 4d. The suitable diameter of Cu MWs strongly affects high-performance devices; this primarily depends on the mechanical strength of the Cu MWs used, which is depends on their diameters50. We measured the Young’s modulus of the electrodeposited Cu microwires using the well-known electric-field-induced resonance method51,
Volatile switching behavior of optimized switches
The highly sensitive dependence of the pull-in voltage on the gap distance can be further exploited to produce a vertical volatile switch with a sub-5 V switching voltage, which may ultimately be useful even in high-density, integrated logic devices. We took advantage of the direct-writing capability of the proposed fabrication technique to fabricate such a 3D vertical switch with a gap distance at the nanometre scale. Figure 6 shows such a vertical switch with a gap distance of ~100 nm that was created via the direct-writing method with an additional slant segment to reduce the distance between the tips of the microwire pair. This vertical switch was found to establish an ‘on state’ at a relatively small bias voltage of ~4.5 V in the first run, as shown in Figure 6c. After removing the applied bias voltage, the microwires deflected immediately back to their original position, which was indicative classic volatile switching behavior. A relatively large contact resistance of ~650 MΩ was found between the microwires that made electric contact. The existence of such a large contact resistance may also be responsible for maintaining a finite voltage drop across the contact and thus a potential difference between the Cu MWs. Therefore, even with electric contact between the microwires, a finite electrostatic force between the microwires, albeit marginally reduced from that before making the electric contact, still remained. This may explain why the microwires deflected back to their original position only when the bias voltage is fully removed instead of deflecting back at the instant of making electric contact, as in typical switching devices with a low contact resistance. In the following switching operations, the pull-in voltage was found to fluctuate due to certain nanoscopic structural changes near the contact site; the amplitude of the change in the switching voltage was found to be ~0.15 V, as shown in Figure 6c. The volatile switching behavior of this vertical switch can be further improved by first reducing the contact resistance between the microwires by coating metals with low contact resistances, such as Au, Pt, Ni or Co, and by further optimizing the mechanical and structural design of the vertical switch.
Novel MEMS switches with true 3D copper-wire-type electrodes were fabricated using the meniscus-confined 3D electrodeposition method. The performance of these devices was found to depend on the dimensional structure of the electrodes, including the active length, the slant angle, the diameter and the distance between the two Cu MWs. The theoretical analysis indicated that the Young’s modulus of Cu MWs was ~120 GPa. Therefore, Cu MWs with diameters between 0.4 and 1.6 μm were produced via 3D direct writing and exhibited elastic moduli near that of bulk Cu. By optimizing the electrode structure, both volatile and nonvolatile switching behaviors were obtained. A volatile MEMS switch with a sub-5 V switching voltage was also demonstrated. The low actuation voltage of the proposed MEMS switch can be used in more novel microdevices designs and applications. Further study should investigate the response behavior and practical application of such vertical switches with low switching voltages that are fabricated using this direct-writing method.
We thank Dr Hezhu Shao for his excellent technical support of the simulations performed in this study. We also acknowledge the financial support of the National Natural Science Foundation of China (11574331 and 11374311), the Ningbo Science and Technology Bureau (2015B11002, 2015A610017, 2012A610121) and the Jiangsu Key R & D program (BE2015104).