#
3-Axis Fully-Integrated Capacitive Tactile Sensor with Flip-Bonded CMOS on LTCC Interposer^{ †}

^{1}

^{2}

^{3}

^{4}

^{*}

^{†}

## Abstract

**:**

## 1. Introduction

## 2. Design

#### 2.1. Sensor Structure

#### 2.2. Working Principle

_{X−}, C

_{X+}, C

_{Y−}, C

_{Y+}, and C

_{Z+}under the diaphragm, and a fixed capacitor C

_{Zref}, which is apart from the other electrodes. When normal force F

_{Z}is applied to the boss of the sensor, capacitances except for C

_{Zref}increase as shown in Figure 3b. In the case that shear force is applied, capacitances correspond to the force direction change. For example, if X-Axis shear force F

_{X}is applied to the sensor in positive direction as shown in Figure 3c, C

_{X+}increases while C

_{X−}decreases. The other capacitances do not change in this case.

_{sens}is the sensor capacitance and the units of f and C

_{sens}are MHz and fF, respectively. The 3-axis output values N

_{X}, N

_{Y}, and N

_{Z}are given by

_{Zref}is the oscillation frequency of C

_{Zref}), and P and f

_{clock}are the count period and the clock frequency, respectively. In this study, P and f

_{clock}are set to be 2

^{13}clock cycles and 1.56 MHz, respectively. When F

_{Z}is applied, N

_{Z}increases, because f

_{Zref}is fixed and f

_{Z+}decreases. In contrast, N

_{X}and N

_{Y}do not change because the decreases of f

_{X−}and f

_{X+}as well as f

_{Y−}and f

_{Y+}are the same in principle due to the symmetry of the electrode layout. If F

_{X}is applied to the positive direction, N

_{X}increases because f

_{X−}increases and f

_{X+}decreases, while N

_{Y}and N

_{Z}do not change because the other oscillation frequencies do not change.

#### 2.3. FEM Simulation

_{Z}and X-Axis shear force F

_{X}of 1 N applied to the positive direction, respectively. The maximum deflections of 3.21 µm for F

_{Z}and 1.03 µm for F

_{X}are smaller than the initial capacitive gap of 4.5 µm. The Z-Axis stiffness was calculated as 3.1 × 10

^{5}N/m. The maximum principal stresses for F

_{Z}and F

_{X}of 1 N were estimated as 365 MPa and 542 MPa, respectively. They are smaller than the reported fracture stress of Si specimens microfabricated by deep reactive ion etching (DRIE) [21]. The eigenfrequencies of the parallel motion of the diaphragm and the rotation motion of the boss were estimated to be 223 kHz and 399 kHz, respectively. These values are much higher than the frequency of the motion of robots and the detectable stimuli frequency of human tactile receptors. Figure 5c,d show the change of 3-axis output values calculated by the simulated 3-axis capacitances as a function of F

_{Z}and F

_{X}, respectively. The nonlinearity of the change of N

_{Z}is caused by the relationship of the oscillation frequency and the capacitance, as described in Equation (1). The sensitivity was estimated as 103 Count/mN for F

_{Z}from 0 N to 0.5 N and 22.6 Count/mN for F

_{X}from −1 N to 1 N, which is high enough to satisfy the minimum force sensitivity of 1 mN. The capacitance at unload condition was calculated as 643 fF for C

_{X−}, C

_{X+}, C

_{Y−}, and C

_{Y+}, 368 fF for C

_{Zref}, and 332 fF for C

_{Z+}.

## 3. Fabrication

_{2}layer was deposited by plasma enhanced chemical vapor deposition (Figure 6b) and the surface was planarized by chemical mechanical polishing. The CMOS substrate was back-ground to make the surface flat and to reduce the thickness to 300 µm (Figure 6c) [22]. After this planarization process, the surface roughness became small enough for the following processes.

_{2}layer was patterned by dry etching for pad opening (Figure 6d). After depositing Au/Pt/Ti seed layers by sputtering, Au of 9 µm thickness was electroplated using a sulfite-based plating solution (Microfab Au 310, Electroplating Engineers of Japan Ltd., Kanagawa, Japan) (Figure 6e). The roles of Au, Pt, and Ti layers are the electrode to electroplate Au, diffusion prevention and adhesion, respectively. The electroplated Au was annealed at 350 °C for 30 min to remove volatile components. After covering the surface of the substrate with a resist for protection, the electroplated Au was planarized by fly-cutting process using a surface planer (DAS8920, DISCO Corporation, Tokyo, Japan) with a diamond bit (Figure 6f) [23]. After removing the resist, the thickness of the electroplated Au was confirmed to be approximately 4.5 µm. The Au/Pt/Ti seed layers were then patterned for forming a ground (GND) electrode and rewiring (Figure 6g). The Au layer was patterned by I

_{2}/KI solution and Pt/Ti layers were patterned by Ar ion milling. The 3-axis capacitor electrodes were formed on an LTCC interposer (Via-Wafer, Nikko Company, Ishikawa, Japan) by the same procedure as stated above (Figure 6h). Figure 8a,b show the optical micrographs of the CMOS substrate and the LTCC interposer substrate just before bonding, respectively.

^{2}. The bonding pads and the interconnections are electrically connected, keeping the isolation of the adjacent bonding pads. A flip-chip bonder (MODEL-6000, HiSOL, Inc., Tokyo, Japan) was used for surface mounting, and a bonding pressure of 3 N was applied at 170 °C. Figure 8e shows a surface-mounted tactile sensor on a glass substrate with interconnections.

## 4. Experiments and Results

#### 4.1. Experimental Method

#### 4.2. Results

_{Z}increased by applying normal force F

_{Z}up to 1.3 N, while the change of N

_{X}and N

_{Y}are small. The sensitivity for F

_{Z}up to 0.5 N is calculated as 34.5 Count/mN. The standard deviations of N

_{Z}at unload condition was 3.5 Count. This corresponds to F

_{Z}of 0.10 mN, which is calculated by the sensitivity of 34.5 Count/mN. The hysteresis for F

_{Z}is less than 0.3% full scale (FS). The nonlinear change of N

_{Z}is due to parasitic capacitance and the nonlinear relationship between the oscillation frequency and the capacitance. Suppose the elastic deformation of the diaphragm, the applied normal force F

_{Z}is given by

_{Z+}is written as

_{Z+}is the area of the Z-Axis sensor capacitor C

_{Z+}, d is the initial gap of the capacitor electrodes, and C

_{p}is parasitic capacitance. Combining Equations (1) and (4)–(6) gives the following equation:

_{Z+,0}is expressed as C

_{Z+,0}= εS

_{Z+}/d. The Z-Axis output value at unload condition N

_{Z0}is given by

_{Z}(i.e., N

_{Z}− N

_{Z0}) is written as

_{clock}, C

_{Z+,0}, and C

_{p}are MHz, fF, and fF, respectively. Comparing Equation (9) and the equation of the approximate curve in Figure 11a, k and C

_{p}are calculated as 2.9 × 10

^{5}N/m and 0.96 pF, respectively. The calculated stiffness is smaller than the simulated one of 3.1 × 10

^{5}N/m. The possible reason is the dimension error (e.g., the thickness of the diaphragm was smaller than the designed value). The parasitic capacitance is caused by the CMOS circuit (e.g., pads and I/O buffers) and the interconnections for rewiring. The parasitic capacitance of the CMOS circuit is about several hundred fF. Thicker isolation layer between the CMOS circuit and the interconnections for rewiring enables the decreasing of parasitic capacitance. When large normal force over 1.3 N is applied, N

_{Z}becomes stable. This is probably because the capacitor electrodes contact by the force.

_{X}and the Y-Axis output value N

_{Y}were linearly changed by the applied shear force, as shown in Figure 11b,c. The sensitivity is calculated to be 15.1 Count/mN for X-Axis and 14.1 Count/mN for Y-Axis. The standard deviations of N

_{X}and N

_{Y}at unload condition were 12.4 Count and 7.6 Count, respectively. These correspond to F

_{X}of 0.82 mN and F

_{Y}of 0.54 mN. The hysteresis is less than 1.7% FS for X-Axis and 1.1% FS for Y-Axis. The average cross-sensitivity is calculated by the following procedures [27]: (1) the 3-axis output values are converted to measured 3-axis force by the approximation curves shown in Figure 11; and, (2) the average cross-axis sensitivity is calculated as the average of the ratio of the measured cross-axis force to the measured principal-axis force. When F

_{Z}is applied, the average cross-axis sensitivity is 8.2% for X-Axis and 10.4% for Y-Axis. In the case that F

_{X}or F

_{Y}is applied, the average cross-axis sensitivity is less than 10%. The calculation of the stiffness and the parasitic capacitance for X-Axis and Y-Axis is not possible by the procedure as stated above because the Z-Axis output value is determined by a fixed capacitor (i.e., f

_{Zref}can be assumed as a constant frequency) and a variable capacitor, while the output values of X-Axis and Y-Axis are calculated by two variable capacitors. Also, definition of the gap change Δd is difficult for X-Axis and Y-Axis because the gap change is not uniform due to the tilt of the diaphragm (Figure 3c), although the gap of the Z-Axis sensor capacitor changes like a parallel plate type capacitor (Figure 3b). As described in Equations (2)–(4), the output values for each axis is calculated by the difference of the count value of the two capacitors. If the count value of each capacitor can be obtained, estimation of the parasitic capacitances for X-Axis and Y-Axis will be possible.

## 5. Conclusions

## Acknowledgments

## Author Contributions

## Conflicts of Interest

## References

- Kappassov, Z.; Corrales, J.-A.; Perdereau, V. Tactile sensing in dexterous robot hands—Review. Robot. Auton. Syst.
**2015**, 74, 195–220. [Google Scholar] [CrossRef] - Dahiya, R.S.; Mittendorfer, P.; Valle, M.; Cheng, G.; Lumelsky, V.J. Directions toward effective utilization of tactile skin: A review. IEEE Sens. J.
**2013**, 13, 4121–4138. [Google Scholar] [CrossRef] - Kumagai, I.; Kobayashi, K.; Nozawa, S.; Kakiuchi, Y.; Yoshikai, T.; Okada, K.; Inaba, M. Development of a full body multi-axis soft tactile sensor suit for life sized humanoid robot and an algorithm to detect contact states. In Proceedings of the 12th IEEE-RAS International Conference on Humanoid Robots, Osaka, Japan, 29 November–1 December 2012; pp. 526–531. [Google Scholar]
- Schmitz, A.; Maiolino, P.; Maggiali, M.; Natale, L.; Cannata, G.; Metta, G. Methods and technologies for the implementation of large-scale robot tactile sensors. IEEE Trans. Robot.
**2011**, 27, 389–400. [Google Scholar] [CrossRef] - Mukai, T.; Onishi, M.; Odashima, T.; Hirano, S.; Luo, Z. Development of the tactile sensor system of a human-interactive robot “RI-MAN”. IEEE Trans. Robot.
**2008**, 24, 505–512. [Google Scholar] [CrossRef] - Yousef, H.; Boukallel, M.; Althoefer, K. Tactile sensing for dexterous in-hand manipulation in robotics—A review. Sens. Actuators A Phys.
**2011**, 167, 171–187. [Google Scholar] [CrossRef] - Dahiya, R.S.; Metta, G.; Valle, M.; Sandini, G. Tactile sensing—From humans to humanoids. IEEE Trans. Robot.
**2010**, 26, 1–20. [Google Scholar] [CrossRef] - Takahashi, H.; Nakai, A.; Thanh-Vinh, N.; Matsumoto, K.; Shimoyama, I. A triaxial tactile sensor without crosstalk using pairs of piezoresistive beams with sidewall doping. Sens. Actuators A Phys.
**2013**, 199, 43–48. [Google Scholar] [CrossRef] - Sohgawa, M.; Hirashima, D.; Moriguchi, Y.; Uematsu, T.; Mito, W.; Kanashima, T.; Okuyama, M.; Noma, H. Tactile sensor array using microcantilever with nickel-chromium alloy thin film of low temperature coefficient of resistance and its application to slippage detection. Sens. Actuators A Phys.
**2012**, 186, 32–37. [Google Scholar] [CrossRef] - Hwang, E.-S.; Seo, J.-H.; Kim, Y.-J. A polymer-based flexible tactile sensor for both normal and shear load detections and its application for robotics. J. Microelectromech. Syst.
**2007**, 16, 556–563. [Google Scholar] [CrossRef] - Lee, H.-K.; Chung, J.; Chang, S.I.; Yoon, E. Normal and shear force measurement using a flexible polymer tactile sensor with embedded multiple capacitors. J. Microelectromech. Syst.
**2008**, 17, 934–942. [Google Scholar] - Dahiya, R.S.; Adami, A.; Collini, C.; Lorenzelli, L. POSFET tactile sensing arrays using CMOS technology. Sens. Actuators A Phys.
**2013**, 202, 226–232. [Google Scholar] [CrossRef] - Takao, H.; Yawata, M.; Sawada, K.; Ishida, M. A multifunctional integrated silicon tactile imager with arrays of strain and temperature sensors on single crystal silicon diaphragm. Sens. Actuators A Phys.
**2010**, 160, 69–77. [Google Scholar] [CrossRef] - Liu, Y.-C.; Sun, C.-M.; Lin, L.-Y.; Tsai, M.-H.; Fang, W. Development of a CMOS-based capacitive tactile sensor with adjustable sensing range and sensitivity using polymer fill-in. J. Microelectromech. Syst.
**2011**, 20, 119–127. [Google Scholar] [CrossRef] - Asano, S.; Muroyama, M.; Bartley, T.; Kojima, T.; Nakayama, T.; Yamaguchi, U.; Yamada, H.; Nonomura, Y.; Hata, Y.; Funabashi, H.; et al. Surface-mountable capacitive tactile sensors with flipped CMOS-diaphragm on a flexible and stretchable bus line. Sens. Actuators A Phys.
**2016**, 240, 167–176. [Google Scholar] [CrossRef] - Asano, S.; Muroyama, M.; Bartley, T.; Nakayama, T.; Yamaguchi, U.; Yamada, H.; Hata, Y.; Nonomura, Y.; Tanaka, S. 3-axis fully-integrated surface-mountable differential capacitive sensor by CMOS flip-bonding. In Proceedings of the 29th IEEE International Conference on Micro Electro Mechanical Systems (MEMS 2016), Shanghai, China, 24–28 January 2016; pp. 850–853. [Google Scholar]
- Hata, Y.; Suzuki, Y.; Muroyama, M.; Nakayama, T.; Nonomura, Y.; Chand, R.; Hirano, H.; Omura, Y.; Fujiyoshi, M.; Tanaka, S. Fully-integrated, fully-differential 3-axis tactile sensor on platform LSI with TSV-based surface-mountable structure. In Proceedings of the 19th International Conference on Solid-State Sensors, Actuators and Microsystems (TRANSDUCERS), Kaohsiung, Taiwan, 18–22 June 2017; pp. 500–503. [Google Scholar]
- Yokoyama, H.; Sohgawa, M.; Kanashima, T.; Azuma, T.; Okuyama, M.; Noma, H. Fabrication and noise reduction of the miniature tactile sensor using through-silicon-via connection with signal amplifier. Jpn. J. Appl. Phys.
**2013**, 52, 06GL08. [Google Scholar] [CrossRef] - Makihata, M.; Tanaka, S.; Muroyama, M.; Matsuzaki, S.; Yamada, H.; Nakayama, T.; Yamaguchi, U.; Mima, K.; Nonomura, Y.; Fujiyoshi, M.; et al. Integration and packaging technology of MEMS-on-CMOS capacitive tactile sensor for robot application using thick BCB isolation layer and backside-grooved electrical connection. Sens. Actuators A Phys.
**2012**, 188, 103–110. [Google Scholar] [CrossRef] - Tanaka, S.; Mohri, M.; Ogashiwa, T.; Fukushi, H.; Tanaka, K.; Nakamura, D.; Nishimori, T.; Esashi, M. Electrical interconnection in anodic bonding of silicon wafer to LTCC wafer using highly compliant porous bumps made from submicron gold particles. Sens. Actuators A Phys.
**2012**, 188, 198–202. [Google Scholar] [CrossRef] - Chen, K.-S.; Ayon, A.; Spearing, S.M. Controlling and testing the fracture strength of silicon on the mesoscale. J. Am. Ceram. Soc.
**2000**, 83, 1476–1484. [Google Scholar] [CrossRef] - Suzuki, Y.; Fukushi, H.; Muroyama, M.; Hata, Y.; Nakayama, T.; Chand, R.; Hirano, H.; Nonomura, Y.; Funabashi, H.; Tanaka, S. 300 μm deep through silicon via in laser-ablated CMOS multi-project wafer for cost-effective development of integrated MEMS. In Proceedings of the 30th IEEE International Conference on Micro Electro Mechanical Systems (MEMS 2017), Las Vegas, NV, USA, 22–26 January 2017; pp. 744–748. [Google Scholar]
- Al Farisi, M.S.; Hirano, H.; Frömel, J.; Tanaka, S. Wafer-level hermetic thermo-compression bonding using electroplated gold sealing frame planarized by fly-cutting. J. Micromech. Microeng.
**2017**, 27, 015029. [Google Scholar] [CrossRef] - Kim, S.-C.; Kim, Y.-H. Flip chip bonding with anisotropic conductive film (ACF) and nonconductive adhesive (NCA). Curr. Appl. Phys.
**2013**, 13, S14–S25. [Google Scholar] [CrossRef] - Bartley, T.; Tanaka, S.; Nonomura, Y.; Nakayama, T.; Hata, Y.; Muroyama, M. Sensor network serial communication system with high tolerance to timing and topology variations. In Proceedings of the IEEE SENSORS 2015, Busan, Korea, 1–4 November 2015; pp. 362–365. [Google Scholar]
- Bartley, T.; Tanaka, S.; Nonomura, Y.; Nakayama, T.; Muroyama, M. Delay window blind oversampling clock and data recovery algorithm with wide tracking range. In Proceedings of the IEEE International Symposium on Circuits and Systems, Lisbon, Portugal, 24–27 May 2015; pp. 1598–1601. [Google Scholar]
- Yoshiyuki Hata, Y.; Nonomura, Y.; Omura, Y.; Nakayama, T.; Fujiyoshi, M.; Funabashi, H.; Akashi, T.; Muroyama, M.; Tanaka, S. Quad-seesaw-electrode type 3-axis tactile sensor with low nonlinearities and low cross-axis sensitivities. Sens. Actuators A Phys.
**2017**, in press. [Google Scholar]

**Figure 1.**Conceptual figure of the 3-axis integrated tactile sensors on a flexible and stretchable bus line: (

**a**) System diagram; (

**b**) Physical implementation.

**Figure 3.**(

**a**) Electrode layout; Working principles for (

**b**) normal force F

_{Z}and (

**c**) X-Axis shear force F

_{X}.

**Figure 4.**(

**a**) Optical micrograph of the complementary metal-oxide semiconductor (CMOS) substrate; (

**b**) Schematic diagram of the capacitive readout circuit; Time responses of (

**c**) the voltage of the sensor capacitor V

_{C}and (

**d**) the oscillation output voltage V

_{out}.

**Figure 5.**Finite element method (FEM) simulation results: Diaphragm deflection by (

**a**) normal force F

_{Z}of 1 N and (

**b**) X-Axis shear force F

_{X}of 1 N applied to the positive direction; Calculated change of 3-axis output values by (

**c**) F

_{Z}and (

**d**) F

_{X}.

**Figure 6.**Fabrication process of the 3-axis integrated tactile sensor: (

**a**) Received CMOS substrate; (

**b**) Thick SiO

_{2}film deposition; (

**c**) Planarization and back grinding; (

**d**) Pad opening; (

**e**) Au electroplating; (

**f**) Planarization of Au bumps and an Au sealing frame; (

**g**) Formation of a ground (GND) electrode and rewiring; (

**h**) Formation of 3-axis capacitor electrodes and rewiring; (

**i**) Au-Au thermo-compression bonding; (

**j**) Formation of a diaphragm; (

**k**) Formation of bonding pads; (

**l**) Surface mounting on a glass substrate with interconnections by an anisotropic conductive film (ACF).

**Figure 8.**Fabrication results: Optical micrographs of (

**a**) CMOS substrate with Au bumps, an Au sealing frame and a GND electrode and (

**b**) low temperature cofired ceramic (LTCC) interposer substrate with 3-axis sensing electrodes; (

**c**) Front side and (

**d**) back side views of a fabricated tactile sensor; (

**e**) Surface-mounted tactile sensor on a glass substrate with interconnections.

**Figure 9.**(

**a**) Experimental system diagram; (

**b**) Digital signal output waveform from the fabricated 3-axis integrated tactile sensor.

**Figure 11.**Change of output value of the 3-axis integrated tactile sensor by (

**a**) normal force F

_{Z}; (

**b**) shear force F

_{X}and (

**c**) shear force F

_{Y}.

Force | Sensitivity [Count/mN] | Noise [Count] | Hysteresis [%] | Average Cross-Axis Sensitivity [%] | ||
---|---|---|---|---|---|---|

X-Axis | Y-Axis | Z-Axis | ||||

F_{X} (−1 to 1 N) | 15.1 | 12.4 | ≤ 1.7 | - | 5.5 | 9.0 |

F_{Y} (−1 to 1 N) | 14.1 | 7.6 | ≤ 1.1 | 5.2 | - | 7.2 |

F_{Z} (0 to 1.3 N ^{1}) | ≥ 34.5 | 3.5 | ≤ 0.3 | 8.2 | 10.4 | - |

^{1}Z-Axis output value N

_{Z}saturate when normal force of over 1.3 N is applied.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Asano, S.; Muroyama, M.; Nakayama, T.; Hata, Y.; Nonomura, Y.; Tanaka, S. 3-Axis Fully-Integrated Capacitive Tactile Sensor with Flip-Bonded CMOS on LTCC Interposer. *Sensors* **2017**, *17*, 2451.
https://doi.org/10.3390/s17112451

**AMA Style**

Asano S, Muroyama M, Nakayama T, Hata Y, Nonomura Y, Tanaka S. 3-Axis Fully-Integrated Capacitive Tactile Sensor with Flip-Bonded CMOS on LTCC Interposer. *Sensors*. 2017; 17(11):2451.
https://doi.org/10.3390/s17112451

**Chicago/Turabian Style**

Asano, Sho, Masanori Muroyama, Takahiro Nakayama, Yoshiyuki Hata, Yutaka Nonomura, and Shuji Tanaka. 2017. "3-Axis Fully-Integrated Capacitive Tactile Sensor with Flip-Bonded CMOS on LTCC Interposer" *Sensors* 17, no. 11: 2451.
https://doi.org/10.3390/s17112451