- Email: [email protected]
Suspended submicron silicon-beam for high sensitivity piezoresistive force sensing cantilevers
Sensors and Actuators A 186 (2012) 80–85
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators A: Physical journal homepage: www.elsevier.com/locate/sna
Suspended submicron silicon-beam for high sensitivity piezoresistive force sensing cantilevers Jia Wei ∗ , Sabrina Magnani, Pasqualina M. Sarro DIMES, Delft University of Technology, Delft, Netherlands
a r t i c l e
i n f o
Article history: Received 3 October 2011 Received in revised form 12 February 2012 Accepted 15 February 2012 Available online 24 February 2012 Keywords: Piezoresistive sensor Force detection Deep reactive-ion etching Submicron suspended silicon-beam
a b s t r a c t This paper presents a submicron suspended piezoresistive silicon-beam structure as a basic sensing element to replace the conventional piezoresistors, so to improve detection sensitivity. The alternative element benefits from the increase in the stress, locally concentrated on the suspended submicron beam, induced by mechanical loads. This approach allows the enhancement of sensitivity without changing the parameters in the mechanical design. A modified deep reactive-ion etching process is developed to create both the suspended silicon-beam and the main mechanical structure in a single etching sequence. The suspended beam is integrated in a silicon force sensing cantilever. A force sensitivity up to 52.5 V/N is obtained, corresponding to a 120% improvement compared to an equivalent structure with conventional piezoresistors. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Piezoresistive sensors have been widely used in many applications, including force detection [1,2], pressure sensors [3], microphones [4] and accelerometers [5,6], due to their small scale, easy integration and convenient readout method. Recent development towards biological [7], chemical detection [8] and atomic force microscope (AFM) probes [9] has further expanded the application of piezoresistive sensing. Different materials besides single crystal silicon and polycrystalline silicon, such as metal or metal oxide films [10,11] and polymer based material [12], have been utilized according to the application specific requirements. For silicon based devices, piezoresistors are often made from doped silicon layers obtained by either ion-implantation or epitaxy. Other methods, such as evaporation [13], sputtering [11] and even inkjet printing [14] have been reported to deposit various of piezoresistive material. For bulk-machined silicon based piezoresistive sensors, modification of geometry and size of the mechanical structure, i.e. by using longer but narrower beams, or larger but thinner membranes are often required, in order to improve sensitivity [15]. A higher signal-to-noise ratio is so achieved and consequently the minimum detectable force, pressure and acceleration are reduced. However, such modifications cause large variations in the mechanical properties of the structure, such as stiffness and resonant frequency. As
a result, the mechanical structure needs to be redesigned. Additionally, such modifications may encounter technology limitations, and often require a more complex fabrication process. To achieve better sensitivity without modifying the mechanical design, the conventional piezoresistive sensors should be replaced with more sensitive elements, such as silicon nano-wires [16], or by other effects [17]. For example, Robert et al. [18] use silicon sub-micron wires integrated with a micron-level mechanical structure, as basic sensing elements for an accelerometer. However, the fabrication involves a complex process and expensive substrates. An alternative piezoresistive sensing element, which can be easily integrated in a conventional fabrication process for piezoresistive sensors, needs therefore to be developed. In this paper, a suspended submicron silicon-beam is proposed as a new piezoresistive element to improve the sensitivity of piezoresistive devices without the need of mechanical redesign or major fabrication process alteration. The concept and theoretical analysis are introduced. A fabrication process with a modified deep reactive-ion etching (DRIE) sequence is developed to integrate the new element. To validate the concept, the new element is integrated on a force sensing cantilever. A conventional piezoresistor is implemented in the same cantilever for performance comparison.
2. Design and analysis
∗ Corresponding author. Tel.: +31 152781285; fax: +31 152622163. E-mail address: [email protected] (J. Wei). 0924-4247/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2012.02.021
Fig. 1 schematically shows a force sensing cantilever integrated with a conventional piezoresistor (Fig. 1a) and the proposed suspended submicron silicon-beam (Fig. 1b). The silicon-beam is
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
81
Fig. 1. Schematic drawing of a force sensing cantilever with (a) a conventional piezoresistor and (b) a suspended silicon-beam sensor; (c) the stress distribution on their vertical cross-sections.
locally suspended on the top surface of the silicon cantilever. When a fixed load is vertically applied at the tip of the cantilever, a linear distribution of stress along the direction of thickness (y-direction) is induced in the structure. On the other hand, the cavity under the proposed suspended beam gives a non-symmetric geometry along the y-direction, and therefore generates a local enhancement of the mechanical stress inside the suspended beam (Fig. 1c). This enhancement results in an improvement of sensitivity for force detection. As the new element occupies only a small part of the entire structure, the influence on the mechanical properties of the whole system is not significant. The enhancement of the stress on the suspended beam can be estimated with a 1D model. Assuming a uniform distribution of longitudinal stress along the width of the cantilever, the cross-sections of the cantilever along the piezoresistor and suspended beam are shown in Fig. 1c. The geometric parameters of the suspended beam and their symbols are listed in Table 1. The distribution of the longitudinal stress (x-direction) is asymmetric due to the non-symmetric geometry in the vertical cross-section. Assuming this distribution follows a linear relation along the location in y-direction, it can be written as
= k(y − y0 )
(y ∈ [0, t − t1 − t2 ] ∪ [t − t1 , t])
=0
(y ∈ (t − t1 − t2 , t − t1 ))
where k is the gradient of the stress distribution, and y0 is the y location of the neutral plane of the cantilever. The distribution of the stress needs to satisfy the following equations:
⎧ t ⎪ ⎪ dy = 0 ⎨ 0 t ⎪ M ⎪ ⎩ (y − y0 )dy = 0
w
where M is the bending moment applied on the cantilever. The location of the neutral plane and the gradient of the stress can be solved respectively as:
⎧ t1 t2 t − t2 ⎪ ⎨ y0 = 2 + t − t 2 ⎪ k = ⎩
3M/w
y03 + (t − y0 )3 − (t − y0 − t1 )3 + (t − y0 − t1 − t2 )3
The non-uniform distribution of stress can be drawn as a function of y (Fig. 2). For a 30 m thick and 50 m wide cantilever, with t1 = 1 m and t2 = 4 m, under M = 1 × 10−6 Nm bending moment, the distribution of the stress is shown in Fig. 2a (blue solid line). Compared to a conventional cantilever (black dot line) under the same mechanical load, the increase in stress at the top surface of the cantilever is about 50%. To better understand the distribution of the longitudinal stress, 2D finite element analysis (FEA) is used (Fig. 2b). The FEA gives a similar distribution of stress (green dash line in Fig. 2a), with slight
Fig. 2. (a) 1D numerical calculation and FEA simulation of the stress distribution on a cross-section (B-B ) of cantilevers with and without a locally suspended silicon-beam. (b) 2D FEA simulation of the suspended beam on the top surface of a silicon cantilever, showing the local concentration of the stress inside the suspended beam. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
82
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
Table 1 Main geometric parameters. Geometric parameters
Symbol
Geometric parameters
Symbol
Length of the suspended beam Width of suspended beam Width of the opening of the local cavity Width of the cantilever
L w1 w2 w
Thickness of the cantilever Thickness of the suspended beam Depth of the local cavity
t t1 t2
Fig. 3. Increase in local stress in the suspended beam as a function of beam thickness and cavity depth. Fig. 4. Main steps of the fabrication process (cross-section at A-A in Fig. 1). The suspended beam is formed using a combination of DRIE and isotropic silicon etching process.
deviations from the 1D numeric model. This deviation is partly due to the linear assumption used in the calculation. The distribution of stress depends on the geometric parameters, such as the beam thickness (t1 ), depth of the cavity (t2 ) and the thickness of the cantilever (t). Fig. 3 illustrates the influence of t1 and t2 to the relative increase in local stress at the surface of the suspended beam. A higher stress can be obtained with a smaller thickness of the beam and a larger cavity depth. However, the smaller the beam is, the easier the beam can be damaged, and its fabrication in a conventional process is more difficult. The depth of the local cavity needs to be small compared with the thickness of the cantilever otherwise the mechanical properties of the cantilever will be influenced. In our experiment, the thickness of the beam is chosen to be 0.5 m, and the depth of the local cavity is chosen to be approximately 10% of the cantilever thickness, i.e. 2–4 m. In the above analysis, the influence of the non-uniformity at the top surface of the cantilever is not considered. However, the concentration of the stress on the top surface has a similar effect as in the vertical cross-section, and therefore it contributes to the sensitivity improvement as well. In this paper, a silicon force sensing cantilever is chosen to validate and demonstrate the proposed concept of the suspended beam. However, this concept can be applied to other silicon based piezoresistive sensors as well.
devices. The piezoresistors were defined in a 500 nm boron-doped (1.1 × 1018 atoms/cm3 ) epitaxial layer (Fig. 4a) grown on an ntype silicon substrate. All resistors were oriented along the [1 1 0] direction in (0 0 1) plane. After the metallization process, tetramethylammonium hydroxide (TMAH) was used to etch cavities from the wafer backside, thus defining the thickness of the silicon cantilevers (Fig. 4c). To fabricate the suspended sub-micron silicon beam in a simple way, a dedicated etching sequence was used by combining a short DRIE step with an isotropic etching step as shown in Fig. 4d. The process parameters are listed in Table 2. The sequence consisted of three steps. 1. A short DRIE step (Recipe A) was performed in the beginning of the sequence to roughly define the thickness of the suspended beam. 2. A passivation layer from C4 F8 was deposited (Recipe B) on the sidewalls of the beam for protection during the following etch step. 3. A silicon isotropic etching step (Recipe C) was then used to create the undercut below the beam to form the local cavity. In this step, the bottom surface of the beam was etched as well. To compensate this loss in beam thickness, the time of the DRIE in the first step was increased accordingly.
3. Fabrication The lateral dimension of the submicron silicon-beam was defined by lithography, and the thickness of the beam and the depth of the cavity were controlled by the time of the etching steps.
An IC-compatible process [19], based on a typical fabrication flow for piezoresistive sensors [20], was used to fabricate the Table 2 Main parameters of the used recipes. Recipe
Steps
SF6 /C4 F8 (sccm)
Energy (W)
Timea (s)
T (◦ C)
Step time (s)
A
Passivation Etching
0/280 400/0
1500 1500
2 5
−10
28
B C
Passivation Etching
0/280 400/0
1500 1500
−10 −10
12 40
a
The switching time of DRIE etching.
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
83
Fig. 5. SEM pictures of (a) fabricated suspended sub-micron silicon beams and (b) close-up of a suspended beam.
Fig. 6. SEM photos of (a) a fabricated force sensing cantilever, (b and c) close-up showing the integrated suspended sub-micron silicon-beam sensors and conventional piezoresistors.
Fabricated suspended beams are shown in Fig. 5. A cross-section of a typical beam is shown in Fig. 5b. The beam was 5 m long, 500 nm wide and 500 nm thick, with a ∼3 m deep cavity. After the definition of the suspended beam, a DRIE process was used to define the geometry of the cantilever with the same recipe (Recipe A) used in the first step of the beam etching (Fig. 4c). Since the suspended beam and the cantilever were etched in the same DRIE equipment, the two processes can be performed in a continuous sequence. In this way, the integration of the suspended sub-micron silicon beam into conventional silicon mechanical structure was achieved without complex process modifications. Fig. 6 shows SEM images of a fabricated force sensing cantilever. The cantilever was 1200 m long, 50 m wide and 35 m thick. Two pairs of piezoresistive sensing elements, conventional piezoresistors and suspended sub-micron silicon beams, were placed on
the top surface of the cantilever (Fig. 6b). The conventional piezoresistors were located close to the middle of the cantilever, and the suspended ones were place close to the edges of the cantilever as show in Fig. 6c. Both elements were 5 m long, 500 nm wide and around 500 nm thick. The cavity under the suspended beam was ∼3 m deep. 4. Results and discussion To validate the proposed concept, first the stiffness measurement of the silicon cantilever was carried out. A force sensing probe (FemtoTools force sensor FT-S270) was used to bend the cantilever downward at its free-end while monitoring the force load (Fig. 7a). This displacement was controlled by a high precision position stage. The force-displacement relation of the cantilever with suspended
Fig. 7. The stiffness measurement of the fabricated cantilever. (a) A photo of the fabricated cantilever during the measurement; (b) measured force versus applied displacement at the tip of the cantilever. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
84
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
silicon-beam is shown in Fig. 7b (green solid line). The extracted stiffness was 48.1 N/m, which is in excellent agreement with the value obtained from a 3D FEA calculation (Comsol). For comparison, the stiffness of a cantilever with the same size but without the suspended element is calculated with Comsol as well (blue dash-line in Fig. 7b). No significant stiffness variation due to the introduced local cavity under the suspended beam was observed, proving that the introduced mechanical modification is negligible compared to the size of the cantilever. To investigate the improvement of sensitivity, the resistance variation from both elements were recorded under same mechanical load (bending down the cantilever) at the tip. Fig. 8 reports the resistance variation of the suspended silicon-beam as a function of the applied force. Both elements show a linear response for a resistance change lower than 15%. The relative variation of resistance in the suspended beam was 120% larger than the variation of the conventional piezoresistor under the same mechanical load. The corresponding sensitivity of the suspended beam in a Wheatstone bridge configuration with 1 V DC supply was 52.5 V/N compared to 23.7 V/N for the conventional one. The theoretical thermal noise of each silicon-beam (typical resistance = 20 k) was 0.02 V/Hz1/2 , corresponding to a force resolution of 0.38 nN/Hz1/2 . The resistance variation in the suspended beam was influenced as well by the local geometry on the horizontal top surface. For example, as illustrated in Fig. 8, the sensitivity of the silicon-beam changed with the width of the opening (w2 in Fig. 1) used for cavity etching. This was partly due to the non-uniform local geometry in the horizontal direction, which causes an extra stress concentration in the suspended beam between the two openings as a function of the opening size. Another cause of the sensitivity variation was due to the depth of the cavity being a bit shallower for a smaller opening size of the etching. Consequently, the effect of the stress concentration was reduced. A non-linear relation between the relative variation of resistance (R/R) and mechanical load was observed under large bending. The relative variation of resistance in the suspended beam began to saturate. This is partly due to the piezoresistive property of the single crystal silicon under large strain. Another cause of the saturation is the fact that, under large mechanical bending of the cantilever, the suspended beam tends to move toward the local cavity under it rather than to follow the curvature of the top surface of the cantilever. This implies that the elongation of the suspended beam is somewhat reduced. The measurement results are supported by a 3D simulation performed using the same geometry sizes as the fabricated devices, in order to take into account the phenomena of the stress concentration in the horizontal direction at the top surface of the cantilever. However, due to the complexity in building the structure model, a cavity with a rectangular cross-section, instead of the exact cavity shape as shown in Fig. 5b, was used under the suspended beam during the simulation. The simulated distribution of the stress, under a 0.5 mN force load vertically applied at the tip of the cantilever, is shown in Fig. 9 (blue solid line). The increase in stress in the suspended beam was 110% of the value for conventional piezoresistor, a slightly lower value than the measurement results. The 2D simulation result (green dash line) with slight deviations from the 3D ones is also shown in the figure. By using the piezoresistive coefficient of single crystal silicon extracted in Ref. [21], where the same fabrication parameters were used, the simulated variation of resistance as a function of applied force is derived (Fig. 10) and again compared with the measurements results. The deviation between the simulated value and the measurement was less than 10% within a 10% variation of resistance value. The deviation was partly due to the inaccurate structure model used in the simulation, and partly due to the non-ideal
Fig. 8. The response of the suspended silicon-beam (for different size of slot w2 ) and the conventional piezoresistor under same mechanical loads applied at the free-end of the sensing cantilever.
Fig. 9. Simulated stress distribution at the B-B cross-section (defined in Fig. 2) of the cantilever for the suspended beam and conventional piezoresistor. A 3D model with the same geometry size used in the fabricated device was used in the simulation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Fig. 10. The measured and simulated resistance variation of the suspended siliconbeam (w2 = 1 m) and the conventional piezoresistor under same mechanical loads applied at the free-end of the sensing cantilever. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
geometry of the fabricated structures. Despite of this slight deviation, the simulation and measured values were consistent with each other, especially on the ratio of the relative resistance variation between two elements, in other words, on the percentage of the sensitivity improvement. This analysis proves that the improvement of the sensitivity is accountable to the increase in local stress concentration obtained by using the proposed suspended beams.
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
5. Conclusions A method to improve the sensitivity of force detection without modifying the mechanical design is proposed. Suspended sub-micron silicon-beams are used to replace conventional piezoresistors. To validate the proposed concept, a force sensing cantilever, where the silicon-beam and the cantilever were realized using a continuous DRIE sequence, was fabricated with an IC-compatible process. With the proposed concept, a 120% improvement of sensitivity of the force sensing cantilever compared with a conventional piezoresistor was measured. This concept and the developed technology can be applied to other silicon based piezoresistive sensors as well. Acknowledgments The authors would like to acknowledge Pablo Estevez from PME group of Delft University of Technology for his help on the stiffness characterization, and the DIMES ICP-group of Delft University of Technology for technical support. References [1] T. Chu Duc, J.F. Creemer, P.M. Sarro, Piezoresistive cantilever beam for force sensing in two dimensions, IEEE J. Sens. 7 (1) (2007) 96–104. [2] D. Viet Dao, T. Toriyama, J. Wells, S. Sugiyama, Silicon piezoresistive six-degree of freedom micro force-moment sensor, Sens. Mater. 15/3 (2003) 113–135. [3] J. Wang, X. Li, A single-wafer-based single-sided bulk-micromachining technique for high-yield and low-cost volume production of pressure sensors, in: 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS’11), 2011, pp. 410–413, Article number 5969491. [4] P.R. Scheepera, A.G.H. van der Donkb, W. Olthui, P. Bergveld, A review of silicon microphones, Sens. Actuators A 44 (1994) 1–11. [5] H.S. Hsieh, H.C. Chang, C.F. Hu, C.L. Cheng, W. Fang, A novel stress isolation guard-ring design for the improvement of a three-axis piezoresistive accelerometer, J. Micromech. Microeng. 21 (2011) 105006. [6] A.M. Fitzgerald, L. Zhang, N.I. Maluf, T.W. Kenn, A high-performance planar piezoresistive accelerometer, J. Microelectromech. Syst. 9 (March (1)) (2000) 58–66. [7] Y. Chen, P. Xu, M. Liu, X. Li, Bio/chemical detection in liquid with self-sensing Pr-oxi-lever (piezo-resistive SiO2 cantilever) sensors, Microelectron. Eng. 87 (2010) 2468–2474. [8] R. Katragadda, Z. Wang, W. Khalid, Y. Li, Y. Xu, Parylene cantilevers integrated with polycrystalline silicon piezoresistors for surface stress sensing, Appl. Phys. Lett. 91 (2007) 083505. [9] A. Boisen, J. Thaysen, H. Jensenius, O. Hansen, Environmental sensors based on micromachined cantilevers with integrated read-out, Ultramicroscopy 82 (2000) 11–16. [10] M. Li, H.X. Tang, M.L. Roukes, Ultrasensitive NEMS-based cantilevers for sensing, scanned probe, and very-high frequency applications, Nat. Nanotechnol. 2 (2007) 114–120. [11] A. Wisitsoraat, V. Patthanasetakul, T. Lomas, A. Tuantranont, Low cost thin film based piezoresistive MEMS tactile sensor, Sens. Actuators A 139 (2007) 17–22.
85
[12] C. Song, A.R. Aiyar, S.H. Kima, M.G. Allen, Exploitation of aeroelastic effects for drift reduction, in an all-polymer air flow sensor, Sens. Actuators A 165 (2011) 66–72. [13] A. Johansson, O. Hansen, J. Hales, A. Boisen, Temperature effects in Au piezoresistors integrated in SU-8 cantilever chips, J. Micromech. Microeng. 16 (2006) 2564–2569. [14] H. Al-Chami, E. Cretu, Inkjet printing of microsensors, in: Proc. IEEE 15th International Mixed-Signals, Sensors, and Systems Test Workshop, 2009, pp. 1–6. [15] A. Loui, F.T. Goericke, T.V. Ratto, J. Lee, B.R. Hart, W.P. King, The effect of piezoresistive microcantilever geometry on cantilever sensitivity during surface stress chemical sensing, Sens. Actuators A 147 (2008) 516–521. [16] L.S. Lim, W.T. Park, L. Lou, H. Feng, P. Singh, Design, fabrication and characterization of ultra miniature piezoresistive pressure sensors for medical implants, Adv. Mater. Res. (2011) 94–98. [17] H.D. Ngo, T. Tekin, T.C. Vu, M. Fritz, W. Kurniawan, B. Mukhopadhyay, A. Kolitsch, M. Schiffer, K. Lang, MEMS sensor with giant piezoresistive effect using metal-semiconductor hybrid structure, in: 16th International Solid-State Sensors, Actuators and Microsystems Conference, TRANSDUCERS’11, 2011, pp. 1018–1021, Article number 5969160. [18] P. Robert, V. Nguyen, S. Hentz, L. Duraffourg, G. Jourdan, J. Arcamone, S. Harrisson, M&NEMS: a new approach for ultra-low cost 3D inertial sensor, Proc. IEEE Sens. (2009) 963–966. [19] J. Wei, S. Magnani, P.M. Sarro, Suspended Submicron silicon-beam for high sensitivity piezoresistive sensing, Proc. Eurosensors XXV (2011) 1437–1440. [20] J. Wei, M. Porta, M. Tichem, U. Staufer, P.M. Sarro, A position and forcedistribution sensor-array for monitoring the contact condition of objects in microhandling, in: Proc. of the 23th IEEE International Conference on Micro Electro Mechanical Systems (MEMS 2010), 2010, pp. 623–626. [21] JiaWei, Silicon MEMS for detection of liquid and solid fronts, Ph.D. Thesis, Delft University of Technology.
Biographies Jia Wei received the Ph.D. Degree in 2010 in microelectronics from Delft University of Technology in Netherlands, and the M.Sc. degree in microelectronics from Tsinghua University in China. His background is 3D micromachining for MEMS sensors and actuators. Since October 2010, he continues his research as a post-do in the ECTM group of DIMES in Delft University of Technology. His recent research topic is integration of sensors and smart optics in 3D wafer level package. Sabrina Magnani was born in Modena, Italy, on September 21, 1986. She received Bachelor degree in Physics from University of Modena and Reggio Emilia, Italy, in 2008. Then, she received Master degree in Physical Engineering from Polytechnic of Turin, Italy, in 2011. The project of her Master thesis was based on the improvement in sensitivity of a cantilever-based piezoresistive force sensor and it has been carried out at DIMES (Delft Institute of Microsystems and Nanoelectronics, Delft University of Technology, Delft, The Netherlands). She is currently holding a research fellowship position in the field of electronic devices at Polytechnic of Turin, Italy. Pasqualina M. Sarro received Laurea degree (cum laude, 1980) in physics, University of Naples, Italy; Ph.D. in Electrical Engineering (1987), Delft University of Technology, the Netherlands, 1987. From 1981 to 1983, she was post-doctoral fellow at Brown University, RI, U.S.A. In 1987, she joined the Delft Institute of Microsystems and Nanoelectronics (DIMES). In 2001 she became Full Professor in Microsystems Technology and in 2009 Chair of the Microelectronics Department. She received the EUROSENSORS Fellow Award (2004), was appointed member of the Royal Netherlands Academy of Sciences (2006) and is IEEE Fellow (2006). She has authored and co-authored more than 400 journal/conference papers on micromachined sensors, actuators and MEMS and is member of technical program committees for several related international conferences.
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators A: Physical journal homepage: www.elsevier.com/locate/sna
Suspended submicron silicon-beam for high sensitivity piezoresistive force sensing cantilevers Jia Wei ∗ , Sabrina Magnani, Pasqualina M. Sarro DIMES, Delft University of Technology, Delft, Netherlands
a r t i c l e
i n f o
Article history: Received 3 October 2011 Received in revised form 12 February 2012 Accepted 15 February 2012 Available online 24 February 2012 Keywords: Piezoresistive sensor Force detection Deep reactive-ion etching Submicron suspended silicon-beam
a b s t r a c t This paper presents a submicron suspended piezoresistive silicon-beam structure as a basic sensing element to replace the conventional piezoresistors, so to improve detection sensitivity. The alternative element benefits from the increase in the stress, locally concentrated on the suspended submicron beam, induced by mechanical loads. This approach allows the enhancement of sensitivity without changing the parameters in the mechanical design. A modified deep reactive-ion etching process is developed to create both the suspended silicon-beam and the main mechanical structure in a single etching sequence. The suspended beam is integrated in a silicon force sensing cantilever. A force sensitivity up to 52.5 V/N is obtained, corresponding to a 120% improvement compared to an equivalent structure with conventional piezoresistors. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Piezoresistive sensors have been widely used in many applications, including force detection [1,2], pressure sensors [3], microphones [4] and accelerometers [5,6], due to their small scale, easy integration and convenient readout method. Recent development towards biological [7], chemical detection [8] and atomic force microscope (AFM) probes [9] has further expanded the application of piezoresistive sensing. Different materials besides single crystal silicon and polycrystalline silicon, such as metal or metal oxide films [10,11] and polymer based material [12], have been utilized according to the application specific requirements. For silicon based devices, piezoresistors are often made from doped silicon layers obtained by either ion-implantation or epitaxy. Other methods, such as evaporation [13], sputtering [11] and even inkjet printing [14] have been reported to deposit various of piezoresistive material. For bulk-machined silicon based piezoresistive sensors, modification of geometry and size of the mechanical structure, i.e. by using longer but narrower beams, or larger but thinner membranes are often required, in order to improve sensitivity [15]. A higher signal-to-noise ratio is so achieved and consequently the minimum detectable force, pressure and acceleration are reduced. However, such modifications cause large variations in the mechanical properties of the structure, such as stiffness and resonant frequency. As
a result, the mechanical structure needs to be redesigned. Additionally, such modifications may encounter technology limitations, and often require a more complex fabrication process. To achieve better sensitivity without modifying the mechanical design, the conventional piezoresistive sensors should be replaced with more sensitive elements, such as silicon nano-wires [16], or by other effects [17]. For example, Robert et al. [18] use silicon sub-micron wires integrated with a micron-level mechanical structure, as basic sensing elements for an accelerometer. However, the fabrication involves a complex process and expensive substrates. An alternative piezoresistive sensing element, which can be easily integrated in a conventional fabrication process for piezoresistive sensors, needs therefore to be developed. In this paper, a suspended submicron silicon-beam is proposed as a new piezoresistive element to improve the sensitivity of piezoresistive devices without the need of mechanical redesign or major fabrication process alteration. The concept and theoretical analysis are introduced. A fabrication process with a modified deep reactive-ion etching (DRIE) sequence is developed to integrate the new element. To validate the concept, the new element is integrated on a force sensing cantilever. A conventional piezoresistor is implemented in the same cantilever for performance comparison.
2. Design and analysis
∗ Corresponding author. Tel.: +31 152781285; fax: +31 152622163. E-mail address: [email protected] (J. Wei). 0924-4247/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2012.02.021
Fig. 1 schematically shows a force sensing cantilever integrated with a conventional piezoresistor (Fig. 1a) and the proposed suspended submicron silicon-beam (Fig. 1b). The silicon-beam is
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
81
Fig. 1. Schematic drawing of a force sensing cantilever with (a) a conventional piezoresistor and (b) a suspended silicon-beam sensor; (c) the stress distribution on their vertical cross-sections.
locally suspended on the top surface of the silicon cantilever. When a fixed load is vertically applied at the tip of the cantilever, a linear distribution of stress along the direction of thickness (y-direction) is induced in the structure. On the other hand, the cavity under the proposed suspended beam gives a non-symmetric geometry along the y-direction, and therefore generates a local enhancement of the mechanical stress inside the suspended beam (Fig. 1c). This enhancement results in an improvement of sensitivity for force detection. As the new element occupies only a small part of the entire structure, the influence on the mechanical properties of the whole system is not significant. The enhancement of the stress on the suspended beam can be estimated with a 1D model. Assuming a uniform distribution of longitudinal stress along the width of the cantilever, the cross-sections of the cantilever along the piezoresistor and suspended beam are shown in Fig. 1c. The geometric parameters of the suspended beam and their symbols are listed in Table 1. The distribution of the longitudinal stress (x-direction) is asymmetric due to the non-symmetric geometry in the vertical cross-section. Assuming this distribution follows a linear relation along the location in y-direction, it can be written as
= k(y − y0 )
(y ∈ [0, t − t1 − t2 ] ∪ [t − t1 , t])
=0
(y ∈ (t − t1 − t2 , t − t1 ))
where k is the gradient of the stress distribution, and y0 is the y location of the neutral plane of the cantilever. The distribution of the stress needs to satisfy the following equations:
⎧ t ⎪ ⎪ dy = 0 ⎨ 0 t ⎪ M ⎪ ⎩ (y − y0 )dy = 0
w
where M is the bending moment applied on the cantilever. The location of the neutral plane and the gradient of the stress can be solved respectively as:
⎧ t1 t2 t − t2 ⎪ ⎨ y0 = 2 + t − t 2 ⎪ k = ⎩
3M/w
y03 + (t − y0 )3 − (t − y0 − t1 )3 + (t − y0 − t1 − t2 )3
The non-uniform distribution of stress can be drawn as a function of y (Fig. 2). For a 30 m thick and 50 m wide cantilever, with t1 = 1 m and t2 = 4 m, under M = 1 × 10−6 Nm bending moment, the distribution of the stress is shown in Fig. 2a (blue solid line). Compared to a conventional cantilever (black dot line) under the same mechanical load, the increase in stress at the top surface of the cantilever is about 50%. To better understand the distribution of the longitudinal stress, 2D finite element analysis (FEA) is used (Fig. 2b). The FEA gives a similar distribution of stress (green dash line in Fig. 2a), with slight
Fig. 2. (a) 1D numerical calculation and FEA simulation of the stress distribution on a cross-section (B-B ) of cantilevers with and without a locally suspended silicon-beam. (b) 2D FEA simulation of the suspended beam on the top surface of a silicon cantilever, showing the local concentration of the stress inside the suspended beam. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
82
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
Table 1 Main geometric parameters. Geometric parameters
Symbol
Geometric parameters
Symbol
Length of the suspended beam Width of suspended beam Width of the opening of the local cavity Width of the cantilever
L w1 w2 w
Thickness of the cantilever Thickness of the suspended beam Depth of the local cavity
t t1 t2
Fig. 3. Increase in local stress in the suspended beam as a function of beam thickness and cavity depth. Fig. 4. Main steps of the fabrication process (cross-section at A-A in Fig. 1). The suspended beam is formed using a combination of DRIE and isotropic silicon etching process.
deviations from the 1D numeric model. This deviation is partly due to the linear assumption used in the calculation. The distribution of stress depends on the geometric parameters, such as the beam thickness (t1 ), depth of the cavity (t2 ) and the thickness of the cantilever (t). Fig. 3 illustrates the influence of t1 and t2 to the relative increase in local stress at the surface of the suspended beam. A higher stress can be obtained with a smaller thickness of the beam and a larger cavity depth. However, the smaller the beam is, the easier the beam can be damaged, and its fabrication in a conventional process is more difficult. The depth of the local cavity needs to be small compared with the thickness of the cantilever otherwise the mechanical properties of the cantilever will be influenced. In our experiment, the thickness of the beam is chosen to be 0.5 m, and the depth of the local cavity is chosen to be approximately 10% of the cantilever thickness, i.e. 2–4 m. In the above analysis, the influence of the non-uniformity at the top surface of the cantilever is not considered. However, the concentration of the stress on the top surface has a similar effect as in the vertical cross-section, and therefore it contributes to the sensitivity improvement as well. In this paper, a silicon force sensing cantilever is chosen to validate and demonstrate the proposed concept of the suspended beam. However, this concept can be applied to other silicon based piezoresistive sensors as well.
devices. The piezoresistors were defined in a 500 nm boron-doped (1.1 × 1018 atoms/cm3 ) epitaxial layer (Fig. 4a) grown on an ntype silicon substrate. All resistors were oriented along the [1 1 0] direction in (0 0 1) plane. After the metallization process, tetramethylammonium hydroxide (TMAH) was used to etch cavities from the wafer backside, thus defining the thickness of the silicon cantilevers (Fig. 4c). To fabricate the suspended sub-micron silicon beam in a simple way, a dedicated etching sequence was used by combining a short DRIE step with an isotropic etching step as shown in Fig. 4d. The process parameters are listed in Table 2. The sequence consisted of three steps. 1. A short DRIE step (Recipe A) was performed in the beginning of the sequence to roughly define the thickness of the suspended beam. 2. A passivation layer from C4 F8 was deposited (Recipe B) on the sidewalls of the beam for protection during the following etch step. 3. A silicon isotropic etching step (Recipe C) was then used to create the undercut below the beam to form the local cavity. In this step, the bottom surface of the beam was etched as well. To compensate this loss in beam thickness, the time of the DRIE in the first step was increased accordingly.
3. Fabrication The lateral dimension of the submicron silicon-beam was defined by lithography, and the thickness of the beam and the depth of the cavity were controlled by the time of the etching steps.
An IC-compatible process [19], based on a typical fabrication flow for piezoresistive sensors [20], was used to fabricate the Table 2 Main parameters of the used recipes. Recipe
Steps
SF6 /C4 F8 (sccm)
Energy (W)
Timea (s)
T (◦ C)
Step time (s)
A
Passivation Etching
0/280 400/0
1500 1500
2 5
−10
28
B C
Passivation Etching
0/280 400/0
1500 1500
−10 −10
12 40
a
The switching time of DRIE etching.
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
83
Fig. 5. SEM pictures of (a) fabricated suspended sub-micron silicon beams and (b) close-up of a suspended beam.
Fig. 6. SEM photos of (a) a fabricated force sensing cantilever, (b and c) close-up showing the integrated suspended sub-micron silicon-beam sensors and conventional piezoresistors.
Fabricated suspended beams are shown in Fig. 5. A cross-section of a typical beam is shown in Fig. 5b. The beam was 5 m long, 500 nm wide and 500 nm thick, with a ∼3 m deep cavity. After the definition of the suspended beam, a DRIE process was used to define the geometry of the cantilever with the same recipe (Recipe A) used in the first step of the beam etching (Fig. 4c). Since the suspended beam and the cantilever were etched in the same DRIE equipment, the two processes can be performed in a continuous sequence. In this way, the integration of the suspended sub-micron silicon beam into conventional silicon mechanical structure was achieved without complex process modifications. Fig. 6 shows SEM images of a fabricated force sensing cantilever. The cantilever was 1200 m long, 50 m wide and 35 m thick. Two pairs of piezoresistive sensing elements, conventional piezoresistors and suspended sub-micron silicon beams, were placed on
the top surface of the cantilever (Fig. 6b). The conventional piezoresistors were located close to the middle of the cantilever, and the suspended ones were place close to the edges of the cantilever as show in Fig. 6c. Both elements were 5 m long, 500 nm wide and around 500 nm thick. The cavity under the suspended beam was ∼3 m deep. 4. Results and discussion To validate the proposed concept, first the stiffness measurement of the silicon cantilever was carried out. A force sensing probe (FemtoTools force sensor FT-S270) was used to bend the cantilever downward at its free-end while monitoring the force load (Fig. 7a). This displacement was controlled by a high precision position stage. The force-displacement relation of the cantilever with suspended
Fig. 7. The stiffness measurement of the fabricated cantilever. (a) A photo of the fabricated cantilever during the measurement; (b) measured force versus applied displacement at the tip of the cantilever. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
84
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
silicon-beam is shown in Fig. 7b (green solid line). The extracted stiffness was 48.1 N/m, which is in excellent agreement with the value obtained from a 3D FEA calculation (Comsol). For comparison, the stiffness of a cantilever with the same size but without the suspended element is calculated with Comsol as well (blue dash-line in Fig. 7b). No significant stiffness variation due to the introduced local cavity under the suspended beam was observed, proving that the introduced mechanical modification is negligible compared to the size of the cantilever. To investigate the improvement of sensitivity, the resistance variation from both elements were recorded under same mechanical load (bending down the cantilever) at the tip. Fig. 8 reports the resistance variation of the suspended silicon-beam as a function of the applied force. Both elements show a linear response for a resistance change lower than 15%. The relative variation of resistance in the suspended beam was 120% larger than the variation of the conventional piezoresistor under the same mechanical load. The corresponding sensitivity of the suspended beam in a Wheatstone bridge configuration with 1 V DC supply was 52.5 V/N compared to 23.7 V/N for the conventional one. The theoretical thermal noise of each silicon-beam (typical resistance = 20 k) was 0.02 V/Hz1/2 , corresponding to a force resolution of 0.38 nN/Hz1/2 . The resistance variation in the suspended beam was influenced as well by the local geometry on the horizontal top surface. For example, as illustrated in Fig. 8, the sensitivity of the silicon-beam changed with the width of the opening (w2 in Fig. 1) used for cavity etching. This was partly due to the non-uniform local geometry in the horizontal direction, which causes an extra stress concentration in the suspended beam between the two openings as a function of the opening size. Another cause of the sensitivity variation was due to the depth of the cavity being a bit shallower for a smaller opening size of the etching. Consequently, the effect of the stress concentration was reduced. A non-linear relation between the relative variation of resistance (R/R) and mechanical load was observed under large bending. The relative variation of resistance in the suspended beam began to saturate. This is partly due to the piezoresistive property of the single crystal silicon under large strain. Another cause of the saturation is the fact that, under large mechanical bending of the cantilever, the suspended beam tends to move toward the local cavity under it rather than to follow the curvature of the top surface of the cantilever. This implies that the elongation of the suspended beam is somewhat reduced. The measurement results are supported by a 3D simulation performed using the same geometry sizes as the fabricated devices, in order to take into account the phenomena of the stress concentration in the horizontal direction at the top surface of the cantilever. However, due to the complexity in building the structure model, a cavity with a rectangular cross-section, instead of the exact cavity shape as shown in Fig. 5b, was used under the suspended beam during the simulation. The simulated distribution of the stress, under a 0.5 mN force load vertically applied at the tip of the cantilever, is shown in Fig. 9 (blue solid line). The increase in stress in the suspended beam was 110% of the value for conventional piezoresistor, a slightly lower value than the measurement results. The 2D simulation result (green dash line) with slight deviations from the 3D ones is also shown in the figure. By using the piezoresistive coefficient of single crystal silicon extracted in Ref. [21], where the same fabrication parameters were used, the simulated variation of resistance as a function of applied force is derived (Fig. 10) and again compared with the measurements results. The deviation between the simulated value and the measurement was less than 10% within a 10% variation of resistance value. The deviation was partly due to the inaccurate structure model used in the simulation, and partly due to the non-ideal
Fig. 8. The response of the suspended silicon-beam (for different size of slot w2 ) and the conventional piezoresistor under same mechanical loads applied at the free-end of the sensing cantilever.
Fig. 9. Simulated stress distribution at the B-B cross-section (defined in Fig. 2) of the cantilever for the suspended beam and conventional piezoresistor. A 3D model with the same geometry size used in the fabricated device was used in the simulation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Fig. 10. The measured and simulated resistance variation of the suspended siliconbeam (w2 = 1 m) and the conventional piezoresistor under same mechanical loads applied at the free-end of the sensing cantilever. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
geometry of the fabricated structures. Despite of this slight deviation, the simulation and measured values were consistent with each other, especially on the ratio of the relative resistance variation between two elements, in other words, on the percentage of the sensitivity improvement. This analysis proves that the improvement of the sensitivity is accountable to the increase in local stress concentration obtained by using the proposed suspended beams.
J. Wei et al. / Sensors and Actuators A 186 (2012) 80–85
5. Conclusions A method to improve the sensitivity of force detection without modifying the mechanical design is proposed. Suspended sub-micron silicon-beams are used to replace conventional piezoresistors. To validate the proposed concept, a force sensing cantilever, where the silicon-beam and the cantilever were realized using a continuous DRIE sequence, was fabricated with an IC-compatible process. With the proposed concept, a 120% improvement of sensitivity of the force sensing cantilever compared with a conventional piezoresistor was measured. This concept and the developed technology can be applied to other silicon based piezoresistive sensors as well. Acknowledgments The authors would like to acknowledge Pablo Estevez from PME group of Delft University of Technology for his help on the stiffness characterization, and the DIMES ICP-group of Delft University of Technology for technical support. References [1] T. Chu Duc, J.F. Creemer, P.M. Sarro, Piezoresistive cantilever beam for force sensing in two dimensions, IEEE J. Sens. 7 (1) (2007) 96–104. [2] D. Viet Dao, T. Toriyama, J. Wells, S. Sugiyama, Silicon piezoresistive six-degree of freedom micro force-moment sensor, Sens. Mater. 15/3 (2003) 113–135. [3] J. Wang, X. Li, A single-wafer-based single-sided bulk-micromachining technique for high-yield and low-cost volume production of pressure sensors, in: 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANSDUCERS’11), 2011, pp. 410–413, Article number 5969491. [4] P.R. Scheepera, A.G.H. van der Donkb, W. Olthui, P. Bergveld, A review of silicon microphones, Sens. Actuators A 44 (1994) 1–11. [5] H.S. Hsieh, H.C. Chang, C.F. Hu, C.L. Cheng, W. Fang, A novel stress isolation guard-ring design for the improvement of a three-axis piezoresistive accelerometer, J. Micromech. Microeng. 21 (2011) 105006. [6] A.M. Fitzgerald, L. Zhang, N.I. Maluf, T.W. Kenn, A high-performance planar piezoresistive accelerometer, J. Microelectromech. Syst. 9 (March (1)) (2000) 58–66. [7] Y. Chen, P. Xu, M. Liu, X. Li, Bio/chemical detection in liquid with self-sensing Pr-oxi-lever (piezo-resistive SiO2 cantilever) sensors, Microelectron. Eng. 87 (2010) 2468–2474. [8] R. Katragadda, Z. Wang, W. Khalid, Y. Li, Y. Xu, Parylene cantilevers integrated with polycrystalline silicon piezoresistors for surface stress sensing, Appl. Phys. Lett. 91 (2007) 083505. [9] A. Boisen, J. Thaysen, H. Jensenius, O. Hansen, Environmental sensors based on micromachined cantilevers with integrated read-out, Ultramicroscopy 82 (2000) 11–16. [10] M. Li, H.X. Tang, M.L. Roukes, Ultrasensitive NEMS-based cantilevers for sensing, scanned probe, and very-high frequency applications, Nat. Nanotechnol. 2 (2007) 114–120. [11] A. Wisitsoraat, V. Patthanasetakul, T. Lomas, A. Tuantranont, Low cost thin film based piezoresistive MEMS tactile sensor, Sens. Actuators A 139 (2007) 17–22.
85
[12] C. Song, A.R. Aiyar, S.H. Kima, M.G. Allen, Exploitation of aeroelastic effects for drift reduction, in an all-polymer air flow sensor, Sens. Actuators A 165 (2011) 66–72. [13] A. Johansson, O. Hansen, J. Hales, A. Boisen, Temperature effects in Au piezoresistors integrated in SU-8 cantilever chips, J. Micromech. Microeng. 16 (2006) 2564–2569. [14] H. Al-Chami, E. Cretu, Inkjet printing of microsensors, in: Proc. IEEE 15th International Mixed-Signals, Sensors, and Systems Test Workshop, 2009, pp. 1–6. [15] A. Loui, F.T. Goericke, T.V. Ratto, J. Lee, B.R. Hart, W.P. King, The effect of piezoresistive microcantilever geometry on cantilever sensitivity during surface stress chemical sensing, Sens. Actuators A 147 (2008) 516–521. [16] L.S. Lim, W.T. Park, L. Lou, H. Feng, P. Singh, Design, fabrication and characterization of ultra miniature piezoresistive pressure sensors for medical implants, Adv. Mater. Res. (2011) 94–98. [17] H.D. Ngo, T. Tekin, T.C. Vu, M. Fritz, W. Kurniawan, B. Mukhopadhyay, A. Kolitsch, M. Schiffer, K. Lang, MEMS sensor with giant piezoresistive effect using metal-semiconductor hybrid structure, in: 16th International Solid-State Sensors, Actuators and Microsystems Conference, TRANSDUCERS’11, 2011, pp. 1018–1021, Article number 5969160. [18] P. Robert, V. Nguyen, S. Hentz, L. Duraffourg, G. Jourdan, J. Arcamone, S. Harrisson, M&NEMS: a new approach for ultra-low cost 3D inertial sensor, Proc. IEEE Sens. (2009) 963–966. [19] J. Wei, S. Magnani, P.M. Sarro, Suspended Submicron silicon-beam for high sensitivity piezoresistive sensing, Proc. Eurosensors XXV (2011) 1437–1440. [20] J. Wei, M. Porta, M. Tichem, U. Staufer, P.M. Sarro, A position and forcedistribution sensor-array for monitoring the contact condition of objects in microhandling, in: Proc. of the 23th IEEE International Conference on Micro Electro Mechanical Systems (MEMS 2010), 2010, pp. 623–626. [21] JiaWei, Silicon MEMS for detection of liquid and solid fronts, Ph.D. Thesis, Delft University of Technology.
Biographies Jia Wei received the Ph.D. Degree in 2010 in microelectronics from Delft University of Technology in Netherlands, and the M.Sc. degree in microelectronics from Tsinghua University in China. His background is 3D micromachining for MEMS sensors and actuators. Since October 2010, he continues his research as a post-do in the ECTM group of DIMES in Delft University of Technology. His recent research topic is integration of sensors and smart optics in 3D wafer level package. Sabrina Magnani was born in Modena, Italy, on September 21, 1986. She received Bachelor degree in Physics from University of Modena and Reggio Emilia, Italy, in 2008. Then, she received Master degree in Physical Engineering from Polytechnic of Turin, Italy, in 2011. The project of her Master thesis was based on the improvement in sensitivity of a cantilever-based piezoresistive force sensor and it has been carried out at DIMES (Delft Institute of Microsystems and Nanoelectronics, Delft University of Technology, Delft, The Netherlands). She is currently holding a research fellowship position in the field of electronic devices at Polytechnic of Turin, Italy. Pasqualina M. Sarro received Laurea degree (cum laude, 1980) in physics, University of Naples, Italy; Ph.D. in Electrical Engineering (1987), Delft University of Technology, the Netherlands, 1987. From 1981 to 1983, she was post-doctoral fellow at Brown University, RI, U.S.A. In 1987, she joined the Delft Institute of Microsystems and Nanoelectronics (DIMES). In 2001 she became Full Professor in Microsystems Technology and in 2009 Chair of the Microelectronics Department. She received the EUROSENSORS Fellow Award (2004), was appointed member of the Royal Netherlands Academy of Sciences (2006) and is IEEE Fellow (2006). She has authored and co-authored more than 400 journal/conference papers on micromachined sensors, actuators and MEMS and is member of technical program committees for several related international conferences.