Elsevier

Wear

Volume 260, Issue 6, 10 March 2006, Pages 580-593
Wear

In situ wear studies of surface micromachined interfaces subject to controlled loading

https://doi.org/10.1016/j.wear.2005.02.070Get rights and content

Abstract

Friction and wear are major limiting factors for the development and commercial implementation of devices fabricated by surface micromachining techniques. These tribological properties are studied using a polycrystalline silicon nanotractor device, which provides abundant, quantitative information about friction and wear at an actual microelectromechanical system (MEMS) interface. This in situ approach to measuring tribological properties of MEMS, combined with high-resolution atomic force microscope (AFM) images of wear tracks, provides insight into the effects of different MEMS surface processing on wear. In particular, monolayer coatings have a significant positive effect, while surface texturing does not strongly affect performance.

Introduction

Microelectromechanical systems (MEMS) have received much attention in recent years, due to their promise as the mechanical counterpart to integrated circuit technologies. Success of MEMS would allow for substantial technological advances in the automotive and telecommunications industries and in the area of national defense. Devices with no contacting components, such as airbag accelerometers [1] and pressure sensors [2], have been successfully marketed. Micromirror arrays for overhead projectors [3], [4] are the only commercialized devices to date that allow adhesive contact between components. Many MEMS device concepts, such as microengines and their associated components, including gears, guides, linear racks, and pop-up mirrors [5], [6], torsional ratcheting actuators [7], wedge stepper motors [8], and stepper motors [9], require not only contact between surfaces, but relative sliding of these surfaces.

MEMS devices are commonly fabricated by a process known as surface micromachining, which leverages the integrated circuit fabrication toolset. Prototype devices consisting of five structural levels can be constructed [10], and commercial devices can also be produced in high volume applications [1]. Polycrystalline silicon (polysilicon) is often used as the structural material in surface micromachining because its material properties have been optimized in many respects. For example, very low residual stress levels (<10 MPa) [11], [12] and low stress gradients (<0.2 MPa/μm) [11], [13] are routinely achieved. Also, although polysilicon is brittle with a microstructure-independent fracture toughness of 1.0 MPa m1/2 [14], it exhibits high strength (∼2–5 GPa) [14], [15]. In practice, strength and fatigue-related failures are not observed because flexure is due to bending stresses in very thin and narrow beams, resulting in very low applied stresses compared to the fracture strength. For example, billions of cycles without failure have been demonstrated in an optical switch application [16]. However, because of high surface-to-volume ratios at the microscale compared to the macroscale, tribological issues including adhesion, friction, and wear become critical.

These issues have been highlighted in several different investigations. One of the earliest surface-micromachined actuators was a microturbine driven by gas flow from a micropipette [17], which highlighted the severe tribology issues that can occur at this scale. After fabrication, the sacrificial material around the structural material was removed by a wet chemical etchant, which was then displaced by water. The devices were air-dried and the associated capillary forces would cause the surfaces to adhere strongly. These devices could be freed by mechanical probing, and they would then work for a limited time. A dynamic friction coefficient of 0.28 was estimated by a deceleration measurement. Operation was limited to about one million cycles at 5000 revolutions per second (∼0.3 m/s with a 20 μm diameter hub) due to wear as limited by run-out of the hub (under unknown loading conditions) [18].

An electrostatic micromotor, driven by phased voltages applied at stators, was also developed at about the same time [19]. From the equations of motion (with the positional dependence of torque calculated from electrostatic simulation) and by comparing to experimental results from video microscopy, the dynamic friction coefficient of the polysilicon in contact with silicon nitride was estimated at between 0.2 and 0.38 [20]. This motor endured for up to 1 min at five revolutions per second before failure.

In these micromotor studies, the apparent applied pressures are not known because small tilts in the device change the area of contact from two parallel surfaces to a line contact with much higher pressures. Lim et al. [21] developed a friction test structure based on a single electrostatic comb drive and a friction foot to which electrostatic voltages could be applied to controllably vary the load. Even though the rubbing surfaces were nominally the same as in the wobble motor, a static coefficient of friction of approximately 5 was measured.

Beerschwinger and colleagues developed specimen-on-disc samples with controlled areas of contact [22]. They coated these samples with various micromachining-compatible materials and measured the friction and wear properties at the same pressures as those calculated in their micromotors. They found that for low nominal contact pressures the wear rate settled into a constant value independent of pressure, but for higher pressures wear rates were similar to that predicted for macroscale contacts. Although their motors did not function well, their calculations, based on measured friction values, indicated the addition of a bushing and a more symmetric stator design would result in working devices [23].

Improved processing techniques to address these tribology issues have been reported. A critical point drying process, which avoids the capillary problem after release, has been developed [24]. Also, processes for depositing organic monolayer lubricants by liquid [25], [26] and vapor [27] deposition routes have been developed. These processes all result in much higher yield of micromachined devices, while the monolayers reduce the friction coefficient to as low as ∼0.1.

In a significant enhancement to the micromotor concept, Sniegowski and Garcia [28] reported a design and process for a microengine that allowed a small gear driven by orthogonal electrostatic comb drives to be rotated via a pin linkage and to couple to other gears. Miller et al. [29] derived the microengine equations of motion and demonstrated that over one billion cycles could be achieved before failure if optimized signals were applied. On the other hand, by operating the device near resonance (3 kHz), Tanner and Dugger found that the device could fail in as few as 100,000 cycles [30]. In their study, microengines and friction test devices were fabricated, and the lifetime and wear of the monolayer-coated microengines compared to friction and wear of the test devices [30]. The failure analysis showed that the debris generated by both devices was an agglomerated silicon oxide. However, much more debris was generated at a low humidity (1%) in the microengine than in the friction test structure, whereas at high humidity (40%) more debris was generated in the test structure than in the microengine. These differences were attributed to differences in the operating speeds, operating pressures and the degree of debris trapping of the two devices.

Indentation testing on silicon [31], as well as scratch testing [32] and atomic force microscopy (AFM) testing [33] on silicon, doped and undoped polysilicon and silicon carbide have been reported. While these studies were not performed on micromachined devices and therefore do not necessarily reflect details of the processing or actual loading conditions, they do give information on the material deformation mechanisms. It has been demonstrated that silicon deforms plastically by undergoing transitions to crystalline and amorphous phases [31]. Similarly, n+ polysilicon appears to deform in a ductile fashion under scratching conditions [32], [33]. This material behavior is possible because high shear stresses can develop in the hydrostatic compressive field under the indenter and scratch tips.

In both the micromotors and the microengine, and also in many other devices where the design allows contacting surfaces, friction is a detrimental effect. Also, it is difficult to know the pressure distribution in detail in these devices. In contrast, the nanotractor1 design studied here takes advantage of friction to obtain high-performance characteristics such as nanometer-scale step size for precision alignment, a large in-plane actuation force (0.5 mN) and a large travel distance (±100 μm) [34]. Furthermore, the same actuator is also a model friction test structure with well-known normal loads from 1 μN [35] to 10 mN [34]. Changes in friction and wear must be understood to ensure that the actuator performance is predictable. Because the actuator and the test structure are integrated into the same device, it is more likely that friction and wear measurements will lead to an understanding of its failure mechanisms than was possible in the studies reported above, since the measurements are taken during actual nanotractor operation.

Based on these considerations, we report here an initial wear study of wear in the nanotractor. Information about induced surface wear is obtained using high-resolution AFM images of worn polysilicon surfaces, which show that wear can be produced in a controlled fashion. Our results also indicate that wear performance is strongly improved by the presence of a monolayer coating while at the same time, it is not yet clear whether surface roughness has any effect on the tribological behavior of these surfaces.

Section snippets

Background information on the nanotractor

The nanotractor is a recently-developed tool for understanding friction and wear for MEMS devices [34]. Its design has been previously discussed in detail [34] and is briefly reviewed here. Fig. 1 shows how motion of the nanotractor is achieved by sequentially actuating and releasing the central driving plate while the clamps are alternately held in place using electrostatic potentials. The actuation plate electrode is segmented by electrically grounded standoffs, as shown in the figure, that

Tests performed

We conducted three types of tests on nanotractors that were manufactured with both load springs A and B connected to them. We refer to the first type as a “wear test”, which is represented schematically in Fig. 5. The sequence of voltages (forces) applied during one cycle of this test, which walks the nanotractor to the right, is shown in Fig. 6. Essentially this is the same as the normal walking sequence in Fig. 1, Fig. 2, but now the leading clamp is subjected to an intentionally applied load

Process variations studied

For this paper, several nanotractor devices were studied, and the roughness and the lubricant coating were varied. These devices and their surface treatments are summarized in Table 1. Devices were selected from two different fabrication lots, “Lot A” and “Lot B”. Four different variants of Lot A were examined. One involved no further surface treatment. Two involved increasing the roughness of the P0 layer by thermally oxidizing the P0 at 900 °C for 100 and 300 min, respectively, promoting uneven

Wear tests of the nanotractor device

Testing occurred within 2 months after the release and coating procedures described above were carried out. Before testing, the devices were stored in air (relative humidity about 30%) inside GelPak™ containers. Environmental conditions during testing were similar (room temperature and relative humidity about 30%). Such conditions could be commonly encountered by a MEMS device.

Data from one representative wear test are shown in Fig. 8 on a Lot A device with CMP post-deposition processing of P0,

Discussion

These preliminary tests provide information about friction and wear properties of polysilicon MEMS surfaces in a well-controlled sliding environment. The data clearly show that performance is greatly enhanced by a monolayer coating. It is not yet clear whether surface roughness has any effect on the tribological behavior of these surfaces. Many more tests need to be performed to achieve statistically significant results on wear lifetime with respect to different surface treatments.

Variations in

Concluding remarks/future work

The work performed thus far indicates that the nanotractor is a promising vehicle for in situ wear studies on MEMS devices. Although more experiments are needed and further development of the testing methodology is required, several conclusions can be made. First, the nanotractor device fails via interfacial seizure due to wear processes at the sliding interfaces under well-characterized loading conditions. Although it is necessary to develop more sensitive in situ tests, this is a necessary

Acknowledgements

We acknowledge the staff at the Microelectronics Development Laboratory at Sandia National Laboratories for fabricating the samples, and in particular Michael J. Shaw for the processing of the Lot A splits. We also acknowledge contributions of Mark D. Street for AFM investigations of wear track topography and determination of wafer roughnesses via AFM, and David S. Grierson for the SEM image of the nanotractor wear track. This work was supported by the US Department of Energy, BES-Materials

References (45)

  • P.F. van Kessel et al.

    MEMS based projection display

    Proc. IEEE

    (1998)
  • J.J. Sniegowski et al.

    Microfabricated actuators and their application to optics

    Miniaturized Systems with Micro-Optics and Micromechanics, San Jose, CA

    (1995)
  • J.J. Sniegowski, S.L. Miller, G.F. LaVigne, M.S. Rodgers, P.J. McWhorter, Monolithic geared mechanisms driven by a...
  • S.M. Barnes et al.

    Torsional ratcheting actuating system

  • J.J. Allen et al.

    Micromachine wedge stepping motor

    ASME IMECE

    (1998)
  • R. Yeh et al.

    Single mask, large force, and large displacement electrostatic linear inchworm motors

    J. Microelectromech. Syst.

    (2002)
  • J.J. Sniegowski et al.

    IC-compatible polysilicon surface micromachining

    Annu. Rev. Mater. Sci.

    (2000)
  • J. Yang et al.

    A new technique for producing large-area as-deposited zero-stress LPCVD polysilicon films: The MultiPoly process

    J. Microelectromech. Syst.

    (2000)
  • M.S. Baker et al.

    Integrated measurement-modeling approaches for evaluating residual stress using micromachined fixed-fixed beams

    J. Microelectromech. Syst.

    (2002)
  • B.D. Jensen et al.

    Interferometry of actuated cantilevers to determine material properties and test structure nonidealities in MEMS

    J. Microelectromech. Syst.

    (2001)
  • R. Ballarini et al.

    Effect of microstructure on the strength and fracture toughness of polysilicon

    Mechanical Properties of Structural Thin Films

    (2001)
  • J. Bagdahn et al.

    Fracture strength of polysilicon at stress concentrations

    J. Microelectromech. Syst.

    (2003)
  • Cited by (47)

    View all citing articles on Scopus
    View full text