Microaccelerometer

In its simplest form, a conventional accelerometer consists of a proof mass, a spring and a position detector. Under steady state conditions, the proof mass experiencing a constant acceleration will move from its rest position to a new position determined by the balance between its mass times the acceleration and the restoring force of the spring. Using a simple mechanical spring, the acceleration will be directly proportional to the distance traversed by the proof mass from its equilibrium position.

 In a force feedback approach, the position of the proof mass is held nearly constant. This is accomplished by feeding back position information to the control electrodes. The resolution of accelerometers is directly proportional to the position detection capability and the effective closed-loop spring constant. Our approach to ultrahigh resolution devices is to incorporate a weak spring with a sensitive position detector[1]. For position sensing, an electron tunneling tip has been suggested[2,3,4]. Tunneling tips have been used in sensors with reported resolution below 0.001Å/(Hz)[5,6,7].

Three distinctly different die are fabricated during the process and these are subsequently assembled using a alloy bonding technique. Electrical contacts are made between layers during the bonding operation. The accelerometer is controlled by electrostatic force plates above and below the proof mass. The lower electrode has a dual role. In operation, it provides a necessary control electrode. When not in operation, it is used to clamp the proof mass and prevents its motion.

 A key element of our design is the placement of the tip such that the spacing between the neutral position of the proof mass and the top of the tip is zero. This has two beneficial effects. First, it reduces sensitivity to off-axis acceleration by eliminating torque. In this design, the springs are composed of beam sections. If the proof mass was not centered, the beams would be deflected. Lateral acceleration of the proof mass would then resolve into a normal and axial loads on the springs. The normal component would deflect the beam and proof mass further, thereby giving a false acceleration reading. In a similar manner, thermal effects due to changes in spring stiffness are also significantly reduced.

 

Layout

 

Figure 1. Schematic Cross-section of a microaccelerometer being fabricated at Northeastern.

Figure 1 shows a cross section of the final device. This drawing is not to scale, but important features have been indicated. The design is based on four die, bonded together to produce the structure shown in the figure. The top die in the figure is referred to as the tip die and has at its center an electron tunneling tip. The tip is approximately 3.75 µm high and with a flat top is shown in Figure 2. The proof mass is assembled from two identical die. This die is comprised of a border region and the proof mass shown in the center of Figure 3. The mass is 1 cm ( 10,000 µm) across. The lowest die is referred to as the force plate die. The proof mass is located in the exact center of the structure and the tunneling tip is designed to touch the proof mass. Figure 4 shows a top view of just the proof mass and springs and Figure 5 and Figure 6 show top views of the tip and force plate dice.

 flat tunneling tip

Figure 2. SEM micrograph of a tunneling tip.

figure 3

 

Figure 3. Layout of the proof mass die.

.... 

 

Figure 4. SEM micrographs showing top view of the proof mass and spring assembly.

Figure 5. Various views of the tip die. The tip is located in the very center of the device. A single shielded electrical lead provides continuity between the tip and its associated wire bond pad. The four quadrant plates surrounding the tip can be used to control pitch and roll. Surrounding the four quadrant plates is a eutectic bond ring.

Figure 6. Various views of the force plate die. The center region contains an oxide covered force plate. A first metal forms the force plate, lead and wire bond pad. Surrounding the center depression and force plate is a eutectic bond ring.

Tip Die

Referring to Figure 5, the tip die is the most complicated structure in the assembly. It not only incorporates the tunneling tip, but contains four field plates to control the pitch and roll of the proof mass as well as its position. The tip die must also accommodate the bonding method use to assemble the four separate die that comprise the accelerometer. Two separate metal layers are required to create the tip die. The first layer is isolated from the second by a 0.5 µm oxide layer. The first metal layer is used for interconnecting the tip and quadrant plates to their associated wire bond pads. We are currently using chromium which survives a subsequent LTO deposition process. This metal layer covers the tip and can be used as the tip metal.

 The second metal layer is Cr/Au. It forms the quadrant plates and covers the electrical bond pads. It also forms a base layer for the eutectic bond ring that surrounds the center of the die. The center region of the die is recessed with respect to the bond ring and bond pads. This is required to obtain spacing between the field plates and the proof mass. When this die is bonded to the proof mass, the perimeter forms a hermetic seal and the bond pads are mechanically bonded and electrically connected to their counterparts on the proof-mass die. This electrical connection scheme is a key feature of the design. By interconnecting the die in the bonding processes, all electrical connections to the completed accelerometer can be made from its front surface.

 To connect the quadrant plates electrically to the bond pads, holes are cut in the oxide layer above the first metal in the vicinity of the quadrant plates. The first metal layer is patterned to connect each quadrant plate electrically to its associated bond pad.

 Finally, a bond metal layer is deposited and patterned on the bond perimeter. The bond perimeter is also recessed, but not as deeply as the quadrant plates. This recess implements a surface referenced bonding technique we are using. The concept is that the width of the bond metal ring is less than the underlying chrome/gold layer, but its thickness is greater than the depth of the recess. When bonding, the eutectic alloy wets both parts and draws them together through surface tension.

 

Proof Mass Die

In Figure 1 the center two dice, shown in cross-section, form the proof mass which is assembled from two identical dice. As a result of the anisotropic etchant used to machine the proof mass, its sides are tapered at 54.7°. The net weight of the proof mass is 0.18 gm. The proof mass is held to the surrounding frame by a set of springs referred to as 'crab legs.' The target thickness of the crab legs is 25 µm. Figure 7 shows a top view of the proof mass and springs. Each individual crab leg is divided into three spring sections, two short ones and one long one. The length of the short sections is 5000 µm but the length of the long spring deviates by 200 µm from 10,000 µm due the requirement to clear the corner ties. The corner ties play no role in the normal motion of the proof mass and do not alter the stiffness of the device in its sensitive axis. These ties are added to increase the stiffness to pitch, roll and yaw [8].The proof mass die is maintained at ground potential and the surface of the proof mass is completely covered by metal. However, at the perimeter of the die, electrical contact between bond pads is routed to facilitate the front surface contacting scheme described above. Figure 7 also shows the metal layer layout for the proof mass. Since the anisotropic etching can take place prior to metallization, some freedom exists in the choice of proof mass metal. Our intention is to provide a complimentary layer to the one used on the tip and force plate chips.

 

Figure 7. Proof mass die detailing springs and metal layer.

Force Plate Die

The force plate die is in many ways similar to the tip die. Again a two metal process is used to provide a bond perimeter which is not electrically connected to the force plate. A contact pad to this perimeter metal is provided. The force plate is fabricated in the first metal which may simply be a single chrome layer similar to that used on the tip wafer. The force plate must be covered with a thick oxide layer (1 µm) which prevents an electrical contact between the proof mass and the force plate when the proof mass is being electrostatically clamped. Choices for the bond perimeter and contact layer (second metal) are identical to those described for the tip wafer. 

Assembly

Figure 1 shows the cross-section of the assembled accelerometer die with the bond pad regions highlighted. In particular, the nature of the bond pad regions is illuminated. Since the bond pads are meant to contact metal lines on the tip die as well as the force plate die, it is necessary to access bond pads on the force plate and tip dice. To the top right of the figure, an etched groove is shown extending down to the top of the first proof mass die. Contacts to the tip and control plates are made at this level. On the extreme left side of the assembled die, an anisotropically etched hole extends down to the force plate wafer. Bond pads for the force plate and its ground are made through openings of this kind. Figure 7 shows the top of the proof mass and the electrical interconnections. Wire bonds are made to one side of each interconnection through the anisotropically etch grooves. The five electrical connections to the tip wafer are made to the opposite end of each interconnection during the eutectic bonding process. Therefore, the pads seen on the tip wafer are connected to the interconnections on the proof mass via the eutectic bond. The opposite side of each interconnection is accessible through an etched groove for wire bonding. The force plate die has only two electrical connections, one to the force plate itself and another to the bond ring. Both are accessible through deep anisotropic grooves.

 

Acceleration Measurements

A simple circuit was constructed for initial testing of the completely assembled accelerometer. It can be shown that the circuit is stable as long as the damping term for the accelerometer is large. This is accomplish by squeeze film damping when operating at one atmosphere. A test bed was assembled which consisted of a spring loaded platform supported by flexures and fitted with a holder for the accelerometer under test. A commercially available accelerometer (Sunstrand 3000) was placed in on the test platform. A piezo-driver was used to move the accelerometer over a distance of about seven microns and at frequencies from 1 Hz to about 200 Hz. Acceleration data from both accelerometers and displacement data from the driver were captured on a storage scope and subsequently printed. Figure 10 shows the response of the tunneling tip accelerometer 53 Hz. Data from the Sunstrand device is also shown. It was collected with a 9 pole active Butterworth filter during the measurement whereas the data for the tunneling tip device was unfiltered. The data for the tunneling tip device shows a 622 Hz signal riding on the 53 Hz response. We believe that this may be a resonance mode of the spring system. A Bode analysis of our force feedback system indicates a cut-off at above 1000 Hz.

Figure 10. Response of the tunneling tip accelerometer in comparison to a Sunstrand 3000.

References

  1. P.M. Zavracky, F. Hartley, N. Sherman, T. Hansen, and K. Warner, "A New Force Balanced Accelerometer using Tunneling Tip Position Sensing," 7th Int. Conf. on Sensors and Actuators, Yokahama, Japan, June 7-10, 1993.
  2. A.A. Braski, T.R. Albrecht, and C.F. Quate, "Tunneling Accelerometer," J. Microscopy, Vol. 152, p. 100, 1988.
  3. S.B. Waltman and W..J. Kaiser, "An Electron Tunneling Sensor." Sensors and Actuators, Vol. 19, pp 201-210, 1989.
  4. M.F. Bocko, "The scanning tunneling microscope as a high-gain, low noise displacement sensor," Rev. Sci. Instrum. Vol. 61, No. 12, December 1990.
  5. T.W. Kenny, W.J. Kaiser, H.K. Rockstad, J.K. Reynolds, J.A. Podosek, and E.C. Vote, "Wide-Bandwidth Electromechanical Actuators for Tunneling Displacement Transducers," J. MEMS, Vol. 3, No. 3, September 1994.
  6. T.W. Kenny, W.J. Kaiser, S.B. Waltman, and J.K. Reynolds, "A Novel Infrared Detector Based on a Tunneling Displacement Transducer," Appl. Phys. Lett, Vol. 59, 1991.
  7. J.J. Yao, S.C. Arney, N.C. MacDonald, "Fabrication of High Frequency Two-dimensional Nanoactuators for Scanned Probe Devices," J. MEMS, Vol. 1, No. 1, pp. 14-22, 1992.
  8. P.M. Zavracky, F. Hartley, and D. Atkins, "New Spring Design and Processes for an Electron Tunneling Tip Accelerometer," to be published.

    9.