Development of packaging for MEMS inertial sensors - IEEE Xplore

Development of MEMS Inertial Sensors Ryszard J. Pryputniewicz, Thomas F. Marinis*, Joseph W. Soucy*, and Cosme Furlong NEST NanoEngineering, Science, and Technology CHSLT Center for Holographic Studies and Laser micro-mechaTronics Mechanical Engineering Department Worcester Polytechnic Institute Worcester, MA 01609 ~

~

* Draper Laboratory Cambridge, MA 02139 Absrracr - Development of MEMS inertial sensors and their packaging for high performance applications requires a balanced approach combining analyses with testing and measurements. There are too many unknown parameters, e.g., material properties, process conditions, and componentslpackage interfaces, to rely solely on analyses during the development. Recent advances in optoelectronic laser interferometric microscope (OELIM) methodology offer a considerable promise for effective testing and measurements to facilitate optimization of packaging for advanced MEMS inertial sensors. Using OELIM methodology, sub-micron deformations of MEMS components and MEMS packages are readily measured with nanometer accuracy and very high spatial resolution over a range of operating conditions. This greatly facilitates characterization of dynamic and thermomechanical behavior of MEMS components, MEMS packages, and other complex material structures. In this paper, the OELIM methodology, which allows remote, noninvasive, full-field-of-view measurements of deformations in near real-time, is presented and its viability for the development of packaging for MEMS inertial sensors is discussed. These discussions are illustrated by representative results that have been obtained relating to improving the packaging for MEMS inertial sensors.

size, reliability, and cost over competing technologies. Furthermore, integrated MEMYCMOS fabrication processes allow development of new sensors for inertial measurements. For example, the development of a very small, lightweight, inexpensive, and rugged Inertial Measurement Unit (IMU) would enable widespread use of GPS-Aided . Inertial Navigation Systems, which would be useful in commercial as well as military applications [4]. To achieve these IMU goals, fabrication processes are being developed that allow monolithic integration of MEMS with driving, controlling, and signal processing electronics. This integration promises to improve performance of the micromechanical devices as well as the cost of fabricating, packaging, and instrumenting these devices with the electronic subsystems in the same manufacturing and packaging process [5-71.

Keywords: MEMS, inertial sensors, packaging, analysis, testing, measurements, nanometer accuracy, optimization.

1. INTRODUCTION

Acceleration-sensing applications, including platform stabilization, are based on the use of inertial sensors, particularly gyroscopes. Invented over 50 years ago [I], these sensors measure parameters of interest with respect to a reference fiame tixed to inertial space. The measurements are of linear acceleration and angular rate and are made by accelerometers and gyroscopes, respectively. Since their invention, major drawback to the commercial use of the inertial guidance systems has been their cost. However, recent advances in microelectromechanical systems (MEMS) technology [2,3], combining semiconductor materials and processing to create integrated microelectronic and micromechanical systems, have led to development of microgyroscopes, Fig. 1, which expect to have advantageous 0-7803-8416-4/04/$20.00 02004 IEEE.

Fig. I,Draper microgyrotest package.

Because of these developments, performance of MEMS inertial sensors has advanced to the point that package induced stresses now limit further improvements in accuracy and stability. The problem of controlling these stresses, or isolating the sensor from them, is compounded by several constraints. The first of these is that the input axis of the sensor must remain accurately aligned to the package in severe vibration and acceleration environments. An equally important constraint is that the sensor must operate over a 56

wide, uncontrolled temperature range in many military and space applications. Optimization of a MEMS sensor package for high performance applications subject to these constraints cannot be accomplished by analysis alone. Rather, use of a new approach to development of MEMS inertial sensors is being made. This approach is based on optoelectronic laser interferometric microscope (OELIM) methodology. In the following sections, a MEMS microgyro used in this study is described, the OELIM methodology is presented, and preliminary results are discussed. 11. THEMICROGYRO

The microgyroscope sensors are usually designed as electronically driven resonators [SI. Such microgyros operate in accordance with the dynamic theory that when an angular rate is applied to a moving body, a Coriolis force is generated. Then, because of the angular rate applied to the axis of the resonating tuning fork, its tines experience a Coriolis force, which produces torsional effects about the axis of the sensor. These effects, which are proportional to the applied angular rate, cause displacements that are measured capacitively in the silicon instrument. Then, the signal sensed is demodulated, amplified, and digitized to form the device output [ 9 ] . A silicon tuning fork microgyro sensor with folded-beam suspension, in which the proof masses (shuttles) are electrostatically driven into resonance with comb drives [SI, is shown in Fig. 2. In this microgyro, rotation is sensed capacitively with respect to the axis in plane of vibration and normal to the direction ofvibration. In the configuration shown in Fig. 2, the microgyro sensor is attached to the substrate through suspension springs. Since the substrate is made of different material than the sensor, any temperature changes, to which the device may be subjected, will induce stresses that could affect the accuracy and reliability of the sensor output. Therefore, it is necessary to quantify the thermal deformations and the effects that the package stresses may have on the device performance, in order to optimize the functional operation of a packaged device.

. .... . . ; I, *" ' Fig. 3. OELIM configuration for study of microgyros: SF is the spatial filter, L1 is the illumination lens, DBS is the direclional beam spliner, MO is the long working distance microscope

1

objective, and PES is the proximal beam splitter.

In the OELlM methodology [IO-121, the CCD camera captures n images, characterized by intensity distributions I , , ( x , y ) , which are defined by the equation

Ill. OELIM METHODOLOGY

OELlM methodology allows highly accurate and precise measurements of absolute shape and differential deformations o f objects under their actual operating conditions. In the OELIM configuration used in this study, Fig. 3, a beam of collimated coherent light is brought into the system and is directed into a spatial filter (SF) assembly consisting of a microscope objective and a pinhole filter. The resulting, expanded, light field is then collimated by lens LI, and redirected by the directional beam splitter (DBS) through the long distance microscope objective lens (MO) to illuminate the object. The proximal beam splitter (PBS) is placed close to the MEMS microgyro. The light reflected from the MEMS being investigated is transmitted back through MO, DBS, and the relay lens to the CCD camera.

+ 0, +

QG>YI1

3

where I , , ( x , y ) and / , ( x , y ) represent intensities of the two interfering beams, i.e., the object beam and the reference beam, Ap(x.y) represe,nts the optical phase difference between the two beams, and 0, is the discrete phase step imposed during recording of the n-th image to facilitate determination of the fringe-locus function n ( x , y ) which relates to displacements and deformations [13]. If n=5, then from 5 relationships of the type of Eq. 1, we obtain

57

The fringe-locus function determined from Eq. 2, in turn, relates to the unknown deformations l,(x,y) via the equation [I31 Fig. 5. OELlM fringe pauems representing deformations of the test samples of Fig. 4 subjected to the same temperature of 71°C: (a) gold bumps, (b) braze, (c)

where K(x,y) is the sensitivity vector representing geometry of the OELIM system. Based on the specific geometry of the OELIM system used to record images of the MEMS being investigated, as accounted for by the sensitivity vector, and the corresponding fringe-locus function, L(x,y) computed from Eq. 3 provides quantitative measurements of deformations of the microgyro under static and dynamic loads [14-17].

mechanical interposer.

IV. TESTSAMPLES In this study, the following three different approaches to sensor packaging were investigated: ( I ) gold bump, (2) gold/tin braze, and (3) mechanical interposer, Fig. 4. r Gold burnos

,-Braze

D

~~r~~~~ ~E

toyhe OELIM fringe pallem of FLg. 5 6 m‘,ximum de’fomation of 1.04 pm: 2D contour representation.

r

fnterooser

A - microavro B - wire bond “I

C - ceramic chip carrier D - lid E - printed circuit board

Fig. 4. Cross Sections of packager showing gold bumps, braze, and interposer attachments o f microgyros.

v.

RESULTS AND DISCUSSION

Fig. 1. TM deformations ofthe braze test sample correspondingto the OELlM fringe patlem of Fig. 5b, maximum deformation of

This section summarizes results obtained using the OELIM methodology when the test samples were either exposed to different temperatures or subjected to dynamic excitation. Representative OELIM fringe patterns, corresponding to thermomechanical (TM) deformations of test samples of Fig. 4, exposed to the temperature of 7 1 T , are shown in Fig. 5. Observation of the fringe patterns of Fig. 5 clearly indicates different responses of the different test samples subjected to the same temperature. These differences can be related to the structural design of the packages for the microgyros in the specific test samples used in this study. Interpretation of the fringe patterns of Fig. 5 yields quantitative results defining TM deformations of the test samples. For example, quantitative analyses of the fringe patterns of Figs 5a and 5b yield results shown in Figs 6 and 7, respectively. The OELIM methodology is also capable of monitoring deformations continuously during the process. That is, as the

2.103 pm: 2D contour representation.

deformations take place, images are recorded in real-time and processed to generate quantitative representation of the corresponding deformation fields. Representative illustration of such results is given in Figs 8 to IO. These figures display deformations of the mechanical interposer (MI) configuration of Fig. 4c. For example, Fig. 8 shows the TM deformations of the MI test sample at 3 5 T , while Figs 9 and I O display T M deformations of the sample at the temperatures of 63°C and 89”C, respectively. Clearly, these figures show changes in the TM deformations as a function oftemperature. It is very instructive to view animation of the experimental results, such as those shown in Figs 8 to IO, corresponding to a specific temperature range of interest, showing T M deformations of the microgyro in 3D as a function of time into the thermal “loading” cycle. These animated displays of 58

Fig. IO. The MI test sample at 89.2"C, maximum deformation of 1.001 pm: (a) 2D contour representation, (b) 3D wireframe representation.

Fig. 8. The MI test sample at 3 8 T , maximum deformation of 0.457 pm: (a) 2D contour representation, (b) 3D wireframe representation.

Based on the results shown in Figs S to IO, maximum TM deformation of the MI test sample is 1.001 Km at the temperature of 89.2"C. Clearly, this deformation is less than half of the deformation of the gold/tin brazed sample at 71T, shown in Fig. 7, but about the same as that of a gold bumps attachment of Fig. 6. OELIM results are also very useful in determinations of shape and deformations of specific features of microgyros, Figs 1 1 and 12.

A

Fig. 9. The MI test sample at 63.8"C. maximum deformation of 0.788 pm: (a) 2D contour representation, (b) 3D wireframe representation. 10

A

I 1 2DD

Fi

the experimental results show deformations in real-time as they take place on.the actual microgyro. The display is over the entire surface ofthe microgyro.

I

am

I

LOO

I II

i

e a

11. Detailed OELIM measurements ofthe posts supporting proof sscs of a microgyro: (a) 2D representation of deformations in the inity of the post, (b) trace along the line A-A of part (a) showing

maximum deformation of40 nm.

59

.......................................

.....1U1......................LI)_UI.II*

time = 0, positionand (b) a later, i.e.,& > 0, position in the vibration cycle of the left proof mass ofthe microg$ro actuated at 10.1 kHz.

400

500

600

700

800

900

1000

,100

r2ao

POSITION. Pleb

Fig. 12. OELIM measurcd mismatch between teeth o f one of drive comb actuators in the microgyro.

Figure 1 1 indicates that deformations across the post, to which folded springs supporting proof masses of the microgyro are attached, amount to 40 nm. These deformations affect positions of the folded springs that are attached to the post which, in turn, influence positions and orientations of the proof masses of the microgyro. Differences in positional suspensions of the proof masses, values of the slopes along the shown edges of these masses, and their overall orientations can be vividly seen from the results shown in Fig. 1 I . Functional operation of microgyros depends on alignment of teeth in the comb drives that actuate the proof masses. This alignment, or rather mismatch between the teeth of the comb drives, can be readily measured using OELlM methodology. Figure 12 displays representative results of such mismatch for one of the comb drives of a microgyro at 2O0C. More specifically, Fig. 12 displays a comparison of deformations along traces traversing the combs of the outside engine and of the proof mass clearly indicating a mismatch between the teeth of the stationary comb and the teeth of the moving comb ranging from 4 0 MI, at the edges of the drive comb, to 210 nm, at the center. A mismatch between teeth of drive combs has adverse effect on performance of microgyros and should be accounted for. The OELlM methodology can also be used to evaluate quality of motion of the proof masses of packaged microgyros operating at their resonance frequencies. This evaluation provides useful information for comparison, e.g., of motions of both proof masses. The OELlM methodology can readily measure motions of the vibrating proof masses using stroboscopic laser illumination of microgyros [ 141. Representative OELIM fringe patterns, corresponding to the inotionsldisplacements of the left and right shuttles of a microgyro driven at 10.1 kHz, are shown in Figs 13 and 14, respectively. Observation of the fringe patterns of Figs 13 and 14 clearly indicates different response of the left and right proof masses at any specific instant. Furthermore, these fringe patterns also

Fig. 14. Representative OELlM fringe panenis for: (a) the initial, i.e., time = 0, position and (b) a later, i.e., time > 0, position in the vibration cycle ofthe right proof mass of the microgyro actuated at 10.1 kHr

show that a given proof mass has motionsldeformations, which are functions oftime in the vibration cycle. This can be related to the structural design and suspension of the proof masses as well as to the way they are actuated by the comb drives, which are acted upon by electrostatic forces during functional operation of the microgyros. lnterpretation of the fringe patterns of Figs 13 and 14 yields quantitative results defining the motionsldisplacements of the proof masses [IS]. The quantitative results show that the left proof mass in its initial position in the vibration cycle, corresponding to the instant of recording of Fig. 13a, has the out-of-plane displacements ranging from -126 to +65 nm, while at the instant corresponding to the recording of Fig. 13b, its displacements range from -147 nm to +70 nm. Similar results can be obtained for the right proof mass indicating that its displacements, for the same temporal positions in the vibration cycle as those used for the descriptions of motions/displacements of the left proof mass, range from -118 nm to +130 nm and from -150 nm to +lo8 nm, respectively. Comparison of these results clearly indicates that the displacement fields not only change as the function of position in the vibration cycle, but also are different for the left proof mass than for the right proof mass at the same instant in the vibration cycle indicating "asymmetry" in motionsldeformations of the proof masses in the dual, or differential, mass microgyroscope considered in this study. Correlation and evaluation of the results presented in this paper were facilitated by the use of ACES methodology [19,20], which merges different solution approaches to obtain answers that would otherwise be difficult to obtain, to improve results, or to validate data using other solution methodologies.

This study was partially supported by the NEST Program at WPI-ME/CHSLT.

VI. CONCLUSIONS A new approach to development of packaging for MEMS inertial sensors was presented. This approach is based on optoelectronic laser interferometric microscope (OELIM) methodology, which allows measurement of sub-micron deformations of microgyroscopes with nanometer accuracy. Using the OELIM methodology, deformations have been measured on the test samples, representing different attachments, i.e., gold bump, goldtin braze, and mechanical interposer, of the sensor to the package. The test samples were subjected to a range of temperatures. Also, proof masses of the packaged microgyros were subjected to dynamic actuation at about IO kHz. The results show that the goldtin brazed tests samples deformed 2.102 pm when subjected to the temperature of 7I0C, while the mechanical interposer test sample deformed only 1.001 pm, when subjected to even higher temperature of 89.2OC. Also, it was shown that deformations of posts, to which folded springs supporting the proof masses of microgyros are attached, may have deformations on the order of 40 nm, depending on the fabrication process used to make the specific microgyros. In additions, the fabrication processes may also lead to mismatch between the teeth of the drive combs actuating the proof masses; this mismatch may be ranging from about -60 nm to +210 nm. It was also shown that motions/displacements of vibrating proof masses of packaged microgyros can be measured during their functional operation using OELIM methodology. These measurements indicate that the out-of-plane components of the displacement vectors can be on the order of 200 nm for the microgyros investigated. The results presented in this paper indicate that the optoelectronic laser interferometric microscope (OELIM) methodology provides quantitative high spatial resolution results with high accuracy and precision for MEMS subjected to either static or dynamic loading conditions. Furthermore, these results are obtained remotely in a noninvasive manner in full-field-of-view and in near real-time. In fact, the OELIM methodology is particularly capable of measurements and characterization of MEMS under their actual operating conditions. As such, the OELIM methodology is a viable tool for development of packaging for MEMS inertial sensors. It should be noted that other microgyros may exhibit different results than those presented in this paper. Therefore, different types of microgyroscopes should be carehlly characterized to determine their specific performance parameters. By understanding the details of structural deformations of different microgyro sensor packages, we can make specific suggestions for improvements in their designs.

[I]

[2] 131

[4]

151

161 [7] [8]

[9]

[IO]

[Ill

[I21

[I31

[I41

[IS] 1161

[I71

ACKNOWLEDGMENTS The test samples used in this study were fabricated at and provided by the C. S. Draper Laboratory, Inc., Cambridge, Massachusetts.

[181

61

REFERENCES M. Weinberg, 1. Bernstein, J. Borenstein, J. Campbell, I. Cousens, B. Cunningham, R. Fields, P. Greiff, B. Hugh, L. Niles, and J. Sohn, “Micromachining inertial instruments: Proc SPIE, 2879:26-36, 1996. R. J. Pryputniewica, “MEMS d e s i y education by case studies,” P o p u No. /MECE200l/DE-23291, A m Soc. Mech. Eng., New York, NY, 2001. R. J. Pryputniewicz, E. Shepherd, I. 1. Allen, and C. Furlong, “Universitv - National Laboratow alliance for MEMS education.” PIOC. l n t e m ~ t . Symp. on MEMi and Nanorechnolo&y (4”-ISMANJ, Charlotte, NC, pp. 364-371,2003, J. J. Allen, R. D. Kinney, J. Sarsfield, M. R. Daily, J. R. Ellis, J. H. Smith. S . Montaeue. - . R. T. Howe. B. E. Bosher. R. Horowitz. A. P. Pismo, A. M. Lemkin, W. Clark, and T. Juneau, “Integrated MicroElectro-Mechanical sensor development for inertial applications,” PLANS98 Position. Lo~ofion,and NmigOrion Symposium, Rancho Mirage, CA, 1998. W. Kuehnel and S . Sherman. “A surface micromachined silicon accelerometer with on-chip detection circuitry,” Sensors ond .4ctuotors A~ 4S.7-36~ 1994 ~~, ~~,~~~ 1. H. Smith, S. Montague, 1. J. Sniegowski, J. Murray, and P. McWhorter, “Embedded micromechanical devices for monolithic integration of MEMS with CMOS,” Proc. IEDM’95, pp. 609-612, 1995. W. C. Tang, T. H. Nguyen, and R. T. H o w , ”Laterally driven polysilicon resonant m i c r o s ~ c N r e s , ”%mors and Actuators, 20:25-32, 1989. N. Barhour and G. Schmidt, “Inertial sensor technology trends,” 7he Draper Technologv Digesl, Draper Laboratory, Cambridge, MA, 3:5-13, 1999. D. Hanson, T. F. Marinis, C. Furlong, and R. J. Pryputniewicz, “Advances in optimization of MEMS inertial sensor packaging,” Proc. Inlerml. Congress on Experimental Mechanics for Emerging Technologies, Sac. Exp. Mech., Bethel, CT, pp. 832-835,2001, G . C. Brown and R. J. Pryputniewicz, “New test methodology for static and dynamic shape measurements of microelectromechanical systems,” Opt. Eng., 39127-136.2000. R. J. Prypumiewicz, C. Furlong, G. C. Brown, E. 1. Pryputniewicz, and M. E. Seta, “Optoelectronic method for shldies of MEMS,” Proc. Inlerml. Congress on Erperimentnl Mechanics for Emerging Technologies, Soc. Exp. Mech., Bethel, CT, pp. 817-820.2001. R. J. Plyputniewicz, C. Furlong, G. C. Brown, and E. 1. Prypumiewicr, ”Optical methodology for static and dynamic measurements of nanadisplacements,” Proc. lntenrat. Congrus on Experimental Mechanics /or Emerging Technologies, Soc. Exp. Mech., Bethel, CT, pp. 826-831,2001, R. J. Prypuuliewicz, “Quantitative determination of displacements and strains from holograms,” Ch. 3 in Springer Series on Optical Sciences, Vol. 68, Holographic inlerfromelp, Rastogi, P. K., Ed., SpringerVerlag, Berlin, pp. 33-72, 1994. R. J. Prypumiewicr, J. J. Allen, C. W. Dyck, and G. C. Brown, “Experimental method far measurements of dynamic characteristics of microgyroscopes,” Pmc. Internal. Symp. on Microscale Systems, Orlando, FL, pp. 84-87,2000. R. J. Pryputniewicz, I. J. Allen, C. W. Dyck, and G. C. Brown, “Dynamic Characterization of microgyros,” Proc. 2 t hAnnuol Symp. ond Exhibition of /MAPS-NE, Boxboro, MA, pp. 218-224,2001, A. R. Klempner, T. F. Marinis, 1. W. Soucy, C. Furlong, and R. 1. Pryputniewicz, “OELIM characterization of thermal deformations of a MEMS package,” Proc. 29* Annual Symp. and Exhibition O/ /MAPSNE, Boxhoro, MA, pp. 149-154.2002. R. J. Pryputniewicz, T. F. Marinis, J. W. Soucy, P. Hefii, and C. furlong, “Novel noninvmive methodology far characterization of packaging for MEMS inertial sensors: Proc 36’ Internat, Symp. 011 Microelectronics, Boston, MA, pp. 392-397,2003. R. J. Pryputniewicz, X. G. Tan, and A. I. Prrekwas, “Modeling and measurements of MEMS gyroscopes,” in press, Proc. PLANS 2004, Monterey, CA, 2004.

[I91 D. R. Pryputniewicr, ACES opproach lo the developmen1 of mm-ocomponenls, MS Thesis, Worcester Polytechnic Institute, Worcester, MA, 1997.

[20] R. J. Pryputniewicz, C. Furlong, G.C. Brown, E. J. Pryputniewicz, and M. Seta, “ACES methodology for studies of MEMS,” Proc. 2s* Annual Symp. andExhibifion of/MAPS-NE, Boxboro, M A , pp. 208-217,ZOOl

62