Sensors and Actuators A 111 (2004) 5762

Strain analysis in MEMS/NEMS structures and devices by using focused ion beam system
Biao Li a, , Xiaosong Tang b , Huimin Xie c , Xin Zhang a
a

Department of Manufacturing Engineering and Fraunhofer USA Center for Manufacturing Innovation, Boston University, Boston, MA 02215, USA b Institute of Material Research and Engineering, Singapore, Singapore c Department of Engineering Mechanics, Tsinghua University, Tsinghua, China Accepted 8 July 2003

Abstract Measurement of residual strain/stress in microstructures using a focused ion beam (FIB) moir technique is demonstrated in this paper. This technique is selected based on advantages of the FIB system in nano-machining, in situ deposition, imaging, and ne adjustment. A nano-grating is directly written on the top of the microstructures by ion milling without any etch mask; the FIB moir pattern is formed by the interference between a prepared specimen grating and raster scan lines. Effects of milling sequence, grating spacing and trench depth on the nano-grating structures and moir fringes have been investigated. Strain evolution in microstructures during underlying sacricial layer etching was studied. The sign of strain was derived from a rotation moir technique. Moreover, a mass loading gauge with nano-gram resolution has been built. Since the local strain of a microstructure itself can be monitored during the process, the FIB moir technique has many potential applications in the mechanical metrology of micro/nano-electro-mechanical-systems (MEMS/NEMS). 2003 Elsevier B.V. All rights reserved.
Keywords: Focused ion beam; Moir; Strain; Mass gauge

1. Introduction Measuring stress/strain in thin lms is especially important in the emerging eld of micro/nano-electro-mechanicalsystems (MEMS/NEMS). A variety of test structures have been designed for residual strain measurement. Sharpe et al. [2] measured strain in poly-silicon lms with the interferometric strain/displacement gage. The relative displacement change generated by strain can be detected from the movement of reective fringes. Guckel and co-workers have utilized beam-buckling method [3] and ring structures [4] for characterizing compressive and tensile strain, respectively. To measure a relatively large residual tensile strain in polyimide lms, an optical microscope inspection of released-beam deformation was shown to be useful [5]. Meanwhile, many kinds of strain measurement devices have been proposed. Pan and Hsu [6] fabricated a strain sensor

This paper was presented at the 16th IEEE MEMS conference, held in Kyoto, Japan, 1923 January 2003 [1], and is an expansion of the abstract as printed in the Technical Digest of this meeting. Corresponding author. Tel.: +1-617-353-1895; fax: +1-617-353-1896. E-mail address: bli@fraunhofer.org (B. Li). 0924-4247/$ see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2003.07.014

comprising of a pair of cantilever beams with different lengths connected by a short tip. The residual strain causes two beams to deect each other, thereby magnifying the deection, which is measured by the tip. Spiral microstructures were used to measure the average strain gradient of thin lms [7]. A bent-beam structure has also been developed for strain measurements [8]. Vernier gauges with double-cantilever microstructures have demonstrated strain sensitivity in the 104 range [9]. Lin et al designed a passive micro strain gauge with a mechanical amplier [10]. A very ne resolution of 105 strain readouts can be achieved for a micro strain gauge with a 500 m-long indicator beam. Recently, focused ion beam (FIB) instruments are becoming prevalent for specimen preparation and inspection in microelectronics and MEMS/NEMS fabrication. This is mainly because FIB system permits microscopic inspection of the sample under consideration before, during, or after the ion-milling process via scanning ion microscopy (SIM) or scanning electron microscopy (SEM). The direct write capability of FIB milling allows nanometer-scale fabrication of specimen grids on the specic region without the requirement of an etch mask [11]. Based on advantages of the FIB system in nano-machining, in situ deposition, imaging, and

58

B. Li et al. / Sensors and Actuators A 111 (2004) 5762

ne adjustment, a FIB moir technique has been proposed by two of the authors to estimate the residual strain of the thin lm [12]. The sensitivity of strain measurement in the FIB moir technique is determined by the grating spacing and the length of the specimen. In this paper, we investigated the impacts of FIB milling sequence, grating spacing and trench depth on the nano-grating structures and moir fringes. The evolution and sign of strain in microstructures has been derived from the FIB moir technique.

2. Experiment 2.1. FIB nano-machining and FIB moir Poly-silicon (Si) microstructures with a poly-Si beam (2 m thick) and a SiO2 sacricial layer (2 m thick) were used for stain analysis [12]. Using a Ga ion beam current of 70 pA nano-machining was carried out at a magnication of 30,000 on the top surface of the poly-Si. A 30 parallel line pattern having identical line width and spacing was created using a computer-generated machining template in the FIB software system. This pattern was subsequently transferred to the cantilever via ion beam milling. The nano-grating is realized by repeating this pattern across the entire length of the cantilever. Fig. 1 shows two sequences for patterning nano-grating structures after FIB milling. In the case of milling from the anchor toward the front (Fig. 1a), the generated charges is accumulated due to residual charges in the prior milling
Fig. 2. SIM images of nano-grating structures on the poly-Si cantilever created by FIB milling (a) from the front toward the anchor and (b) from the anchor toward the front.

area. As a result, the milling pattern is shifted and the border between two sequent patterns is coarse (Fig. 2a). Fig. 1b shows a reverse milling sequence, i.e. patterning from the front toward the anchor. This leads to a ne border between two sequent patterns due to better grounding (Fig. 2b). FIB scanning moir is formed by the interference between the scanning lines of the FIB scan raster (reference grating) and the specimen grating under specic magnications of SIM [12]. For the nano-grating structures with coarse border, several tiny fringes are obvious in the moir pattern as shown in Fig. 3. Such tiny fringes result in errors in strain analysis since they interrupt the periodicity of moir pattern. The moir fringes are clear and uniform for the nano-grating with ne border.

Fig. 1. The sequence of patterning nano-grating structures by FIB nano-machining. (a) Milling from the anchor towards the front and (b) milling from the front towards the anchor.

Fig. 3. FIB moir patterns of nano-gratings (ruling width: 80 nm, spacing: e 160 nm) on the unreleased poly-Si cantilevers with length of 60 and 80 m, respectively. The nano-grating was created by FIB milling from the anchor to the front (60 m long cantilever) and from the front to the anchor (80 m long cantilever), respectively.

B. Li et al. / Sensors and Actuators A 111 (2004) 5762

59

as shallow as 20 nm. Hence the impact of surface damage on the intrinsic strain of microsystems is negligible, leading to versatile applications of moir technique in monitoring the local strain of a particular structure during various stages of fabrication processes. 2.2. Rotation FIB moir The sign of strain can be determined by a rotation moir technique. If the specimen grating is not subjected to elongations or contractions in the direction perpendicular to the grating, but to rotation only, then the phenomenon is one of pure rotation of two like gratings with respect to each other. The light fringes passes through the intersection of two grating lines, AB [13], shown in Fig. 6. As the specimen is strained, compression on the specimen will cause a decrease in the rotational angle of the fringes (line AB ) while tension will cause an increase in fringe rotation (line AB ). The rotation of moir fringes and raster scans follows the same or reverse direction for tensile or compressive strain, respectively. Moreover, tension will cause a gain in the number of fringes, and compression will cause a reduction in the number of fringes per unit length [14]. The FIB system provides the ability of ne tuning the magnication (1) and tilt angle (0.1 ). The horizontal moir fringes can be obtained by precise alignment of the stage to the grating feature. Therefore, the position shift due to sample re-loading can be eliminated. The vertical position shift (in perpendicular with the FIB raster scan lines) has no effect on moir fringes since no change of the pattern spacing was observed after moving the sample stage in a step of half specimen grating spacing. To investigate the stability of the instrument, a standard mica grating was used as a reference during FIB imaging. It is found that the change of FIB raster scan spacing is negligible from day to day. Thus we can investigate the evolution and sign of strain in the microstructure during various stages in the fabrication process. Fig. 7 shows rotation FIB moir fringes of the poly-Si cantilever before and after removing the underlying sacricial layer. The ruling width and spacing of nano-grating is 80 and 160 nm, respectively, and the length of cantilever beam is 60 m. Both samples are rotated counterclockwise at an angle of 5 for moir investigation. As guided by arrows, the shift of moir fringes is obvious for the sample before and after release. The number of fringes per unit length increases for the released poly-Si cantilever beam, indicating the strain is tensile after removing the underlying sacricial layer. 2.3. Strain evolution Strain in microstructures was derived by calculating the average spacing of moir fringes; higher sensitivity in strain measurements was achieved through creating a denser grating on a longer specimen. In this study, a 7100 lines/mm

Fig. 4. FIB machined nano-grating structures on poly-Si with different spacing.

The resolution of FIB nano-machining is limited by the surface roughness and material re-deposition. Fig. 4 shows various nano-grating structures with different spacing on poly-Si after FIB milling. The grain size of poly-Si is 120 nm. The minimum linewidth of rulings that can be distinguished is about 60 nm. FIB moir fringes is related to the depth of milling trench. As shown in Fig. 5, FIB moir can be observed for a trench

Fig. 5. Moir fringes of nano-gratings with different trench depth. The e ruling width and spacing of specimen grating are 70 and 140 nm, respectively.

60

B. Li et al. / Sensors and Actuators A 111 (2004) 5762

FIB scan lines

>0 =0 <0

Specimen grating

Fig. 6. Moir fringes produced by rotation of reference grating at an angle , relative to the specimen grating subjected to compressive ( < 0), zero e ( = 0) and tensile ( > 0) strains, respectively. Lines AB , AB and AB refer to light fringes in FIB moir pattern. e

grating has been successfully created on a poly-Si surface by FIB milling. We investigated the strain evolution in the poly-Si cantilever during the etching of the SiO2 sacricial layer underneath. A 140 nm nano-grating was fabricated on two cantilevers with different widths, 10 and 4 m, respectively, using FIB milling. The length of both cantilevers is 60 m. A FIB moir pattern was recorded after timed etching in HF. Fig. 8 shows the strain evolution of the poly-Si cantilever during the HF etching process. After 1.5 min of etching, 0.8 m LTO sacricial layer is removed. The widths of the remaining SiO2 lying under cantilevers with 4 and 10 m widths are 2.4 and 8.4 m, and the volume ratios of structural to sacricial layers are 1.71 and 1.22, respectively. The tensile strains of 0.02% and 0.08% are derived for the 10 and 4 m width cantilevers, respectively. Larger strain in the 4 m wide cantilever results from the larger poly-Si/LTO volume ratio. After 5 min of etching in HF, the LTO sacricial layer under the 4 m width cantilever is found to be completely removed through SEM investigation. The SiO2 that remained is found next to the anchor of the 10 m width cantilever. The relative strains are 0.336% and 0.21% for 4 and 10 m

width cantilevers, respectively. Further HF etching results in a slight change of strain, implying that the SiO2 layer closer to the front of the cantilever has more contribution to the strain release. It is also found that the relative strain in the 4 m width cantilever is larger than that in the 10 m width cantilever. The average strain ratio of 104 m width cantilevers is 2.33, which corresponds to the width ratio 2.5. This indicates that the larger strain in a narrower structure is caused by its smaller stiffness. 2.4. Mass loading gauge Micro-cantilever sensors have been widely used to detect surface mass loading based on the cantilever deection and resonance frequency variation. The deections of these cantilevers can be detected with sub-angstrom precision using current techniques, perfected for AFM technology, such as optical, piezoresistive, piezoelectric, capacitive, and electron tunneling. We propose a novel detection mode to measure the cantilever deection based on the shift of moir fringes. To increase the sensitivity of cantilever deection to loading mass, the stiffness of the cantilever should be as small as possible. Since the stiffness is related to the geometry, we

Fig. 7. Rotation FIB moir fringes of the poly-Si cantilever (a) before e and (b) after removing the underlying sacricial layer. Arrows are guides to eyes.

Fig. 8. Strain evolution of poly-Si cantilevers during the sacricial layer etching. The top sequences schematically show the volume change of the SiO2 sacricial layer underneath.

B. Li et al. / Sensors and Actuators A 111 (2004) 5762

61

Fig. 9. (a) Schematic showing the mass gauge fabrication steps; (b) FIB imaging of the reference and sample beams; (c) SEM picture of selective Pt deposition and (d) the relationship between the strain and the loaded mass.

cut the released cantilever into two beams after creating the grating lines as shown in Fig. 9a (step 1). One beam is further machined using FIB milling under an oblique angle (45 ) by tilting the sample stage (step 2) [15]. The surface mass loading is realized by depositing platinum (Pt) on the free end of the sampling beam within the FIB system (step 3). Fig. 9b shows the splitting poly-Si cantilever beam. The grating spacing is 140 nm. Since the nano-grating was fabricated before FIB cut, the square beam with high stiffness can be used as a reference while the triangle beam with small stiffness is used as a gauge for the detection of mass loading. The selective deposition of Pt on the sample beam is demonstrated in Fig. 9c.

Fig. 9d represents the relationship between the strain and the loaded mass. Solid line and dots are derived from FIB moir measurement while the dashed line is the simulated in-plane strain from the nite element method. Both experimental and calculated data show that tensile strains occur, which increases as loading mass increases. However, the measured strains are about two times larger than the simulated data that may result from the variation of Pt density during deposition [16]. Moreover, the slope of the measured data is less steep than the simulated prole. One possible reason is that the moir fringes are recorded in a xed xy plane rather than along the neutral axis of a cantilever. The surface mass loading results in a tensile strain and beam

62

B. Li et al. / Sensors and Actuators A 111 (2004) 5762 [3] H. Guckel, T. Randazzo, D.W. Burns, A simple technique for the determination of mechanical strain in thin lms with applications to polysilicon, J. Appl. Phys. 57 (1671) 1985. [4] H. Guckel, D.W. Burns, H.A.C. Tilmans, D.W. DeRoo, C.R. Rutigliano, Mechanical properties of ne grained polysiliconthe repeatability issue, in: Proceedings of the IEEE Solid State Sensor and Actuator Workshop, Hilton Head Island, SC, USA, 1988, p. 96. [5] M. Mehregany, R.T. Howe, S.D. Senturia, Novel microstructure for the in situ measurement of mechanical properties of thin lms, J. Appl. Phys. 62 (1987) 3579. [6] C.S. Pan, W. Hsu, A microstructure for in situ determination of residual strain, IEEE J. Microelectromech. Sys. 8 (1999) 200. [7] L.-S. Fan, R.S. Muller, W. Yun, R.T. Howe, J. Huang, Spiral microstructures for the measurement of average strain gradients in thin lms, in: Proceedings of the IEEE Microelectromechanical Systems Workshop, Napa Valley, CA, USA, 1990, p. 177. [8] Y.B. Gianchandani, K. Naja, Bent-beam strain sensors, IEEE J. Microelectromech. Sys. 5 (1996) 52. [9] C.-J. Kim, R.S. Muller, A.P. Pisano, Residual strain measurement of thin lms using microfabricated Vernier gauges, Sens. Mater. 4 (4) (1993) 291. [10] L. Lin, A. Pisano, R. Howe, A micro strain gauge with mechancial amplier, IEEE J. Microelectromech. Sys. 6 (1997) 313. [11] M. Kuball, F.H. Morrissey, M. Benyoucef, I. Harrison, D. Korakakis, C.T. Foxon, Nano-Fabrication of GaN pillars using focused ion beam etching, Phys. Stat. Sol. (a) 176 (1999) 355. [12] B. Li, H. Xie, B. Xu, R. Geer, J. Castracane, Investigation of strain in microstructures by a novel moir method, IEEE J. Microelectromech. Sys. 11 (2002) 829. [13] A.J. Durelli, V.J. Parks, Moir Analysis of Strain, Prentice-Hall Inc., Englewood Cliffs, NJ, 1967. [14] R. Wolfe, R. Rowlands, C. Lin, Full eld stress/strain analysis: use of moir and TSA for wood structural assemblies, in: Proceedings of the Experimental Mechanics Plenary Session at the Forest Products Research Society Annual Meeting, Portland, ME, 1994, p. 23. [15] R. Puers, S. Reyntjens, D. Bruyker, The Nanopiranian extremely miniaturized pressure sensor fabricated by focused ion beam rapid prototyping, Sens. Actuators A 9798 (2002) 208. [16] T. Tao, J. Ro, J. Melngaillis, Focused ion beam induced deposition of platinum, J. Vac. Sci. Technol. B8 (1990) 1826.

curved out of the xy plane. The tensile stress tends to elongate the specimen grating spacing while beam bending will cause the reduction of specimen grating projected in the xy plane.

3. Concluding remarks Nano-grating structures with 70 nm in linewidth have been created using FIB direct milling without any etching mask. Effects of milling sequence, grating spacing and trench depth on the nano-grating structures and moir fringes have been investigated. FIB moir technique has many advantages in strain analysis of MEMS/NEMS structures and devices. First, moir pattern can be well distinguished in microstructures using nano-gratings due to the capability of ne tuning of the magnication in FIB. Second, strain change at various fabrication stages can be tracked due to the capability of ne adjusting of the stage position in FIB. Third and nally, a nano-mass gauge can be realized based on FIB machining and FIB moir techniques.

References
[1] B. Li, X. Tang, H. Xie, X. Zhang, Focused ion beam (FIB) nanomachining and FIB moire technique for strain analysis in MEMS/NEMS structures and devices, in: Proceedings of the 16th IEEE International Conference on Micro Electro Mechanical Systems (MEMS03), Kyoto, Japan, 2003, p. 674. [2] W.N. Sharpe, B. Yuan, R. Vaidyanathan, Measurement of Youngs modulus, Poissons ration, and tensile strength of polysilicon, in: Proceedings of the 10th IEEE International Workshop on Microelectromechanical Systems (MEMS97), Nagoya, Japan, 1997, p. 424.