A Micro-Force Sensor with Beam-Membrane

micromachines Article

A Micro-Force Sensor with Beam-Membrane Structure for Measurement of Friction Torque in Rotating MEMS Machines Huan Liu 1, *, Zhongliang Yu 2 , Yan Liu 3 and Xudong Fang 3 1 2 3

*

Key Laboratory of Thin Film and Optical Manufacturing Technology, Ministry of Education, Xi’an Technological University, Xi’an 710032, China School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China; [email protected] State Key Laboratory for Manufacturing Systems Engineering, Xi’an Jiaotong University, Xi’an 710049, China; [email protected] (Y.L.); [email protected] (X.F.) Correspondence: [email protected]; Tel.: +86-29-8617-3349

Received: 20 July 2017; Accepted: 10 October 2017; Published: 12 October 2017

Abstract: In this paper, a beam-membrane (BM) sensor for measuring friction torque in micro-electro-mechanical system (MEMS) gas bearings is presented. The proposed sensor measures the force-arm-transformed force using a detecting probe and the piezoresistive effect. This solution incorporates a membrane into a conventional four-beam structure to meet the range requirements for the measurement of both the maximum static friction torque and the kinetic friction torque in rotating MEMS machines, as well as eliminate the problem of low sensitivity with neat membrane structure. A glass wafer is bonded onto the bottom of the sensor chip with a certain gap to protect the sensor when overloaded. The comparisons between the performances of beam-based sensor, membrane-based sensor and BM sensor are conducted by finite element method (FEM), and the final sensor dimensions are also determined. Calibration of the fabricated and packaged device is experimentally performed. The practical verification is also reported in the paper for estimating the friction torque in micro gas bearings by assembling the proposed sensor into a rotary table-based measurement system. The results demonstrate that the proposed force sensor has a potential application in measuring micro friction or force in MEMS machines. Keywords: beam-membrane structure; micro-force sensor; MEMS bearing; friction torque

1. Introduction The micro gas bearing is frequently used as a very important component for rotating micro-electro-mechanical system (MEMS) devices that can carry load in micro-rotating machinery [1]. Compared to macro gas bearings, MEMS-based bearings work at a higher speed and require better repeatability and stability. Current studies mainly investigate the design and fabrication [2,3], and slip effect [4,5] of gas bearings, but little attention is given to the characterization of the friction torque. Investigating the friction torque of gas bearings can help control lubrication flow and propose a more effective design scheme. However, accurate calculation of friction torque is still a challenge [6–8], and performing experimental analysis inevitably brings forth a favorable solution to the investigation of micro-friction torque in MEMS gas bearings. Because of the very small size and target parameters in the gas bearing, conventional methods cannot provide enough accuracy, both in measurement and installation precision [9,10]. Therefore, the authors set up a rotary table-based measurement system to accomplish the task. The friction torque can be transformed into the more easily detected force parameter by a force

Micromachines 2017, 8, 304; doi:10.3390/mi8100304

www.mdpi.com/journal/micromachines

Micromachines 2017, 8, 304

2 of 12

arm, and the rotary table with necessary gas passage are all available in the laboratory. Then the force sensor becomes the key device for the system. As mentioned above, the sensor should possess a large measuring range to enable its usage in the measurement of different friction status (dry or lubricated), a good sensitivity to accurately detect the target force, and an overload protection to avoid unexpected breakage during sensor installation or incorrect experiments. The literature shows that several MEMS force-sensing devices have been developed. The beam structure and membrane structure are typically used in the sensing configuration [11,12]. The membrane sensor features a relatively large measurement range, but its high supporting stiffness inevitably decreases the measurement sensitivity, which places obstacles in the sensor applications; though reducing the thickness of membrane can tackle this problem, thin membrane may increase the difficulty in sensor fabrication and preserving dimensions accuracy. The beam structure is simply constructed with high sensitivity, but the measurement range is limited by the stiffness of the beams and low linearity may appear when the force-induced deflection is large [13]. A. Jordan and S. Büttgenbach [14] utilize SU-8 resistance to develop a membrane force sensor with low bending stiffness, but it is difficult to fabricate and the low admissible stress of resistance may also limit the range for force measurement. G. Palli and S. Pirozzi [15] developed an optical force sensor and it is very effective in measuring the tendon tension, but the inherent complexity makes it hard assemble with the torque measurement system. Yoshikawa et al. [16] reported a nanomechanical membrane-type surface stress sensor which provides high accuracy, but the planar scheme is unsuitable for contact force measurement. This paper aims to provide an available solution to the force sensor with relatively large measurement range (0–40 mN) and favorable sensitivity (0.1 mV/mN), which can be used in the friction torque measurement system. The requirements for the sensors are a measurement range of 40 mN and a sensitivity of about 0.1 mV/mN, considering the to-be-measured force, unexpected collision in the operation and conditions in the laboratory. Focused on these target performances, a piezoresistive force sensor with beam-membrane structure is proposed. Finite element models for the conventional beam structure and membrane structure are established to gain insight into the influence of configuration on the sensor performances and an improved design attempt is based on the analysis results. The sensing structural design, piezoresistors arrangement, probe choosing and overload protection are all conducted in the design section. Then, the fabricated sensor prototype are packaged and calibrated, and a practical friction torque measurement is performed to verify the feasibility of utilizing the developed sensor in the measurement system. 2. Design The schematic diagram of a measurement system for friction torque in the MEMS gas bearings is illustrated in Figure 1. In the system, the to-be-measured torque is transformed into a force by the lever arm fixed onto the shaft of bearings. The force can be sensed by the relevant piezoresistive sensor, the output voltage of which is acquired by the oscillograph building up the variation curve of friction torque during the rotation of the bearings. The excitation voltage for the Wheatstone bridge is provided by a direct current (DC) power supply and the gas supply is used to provide the lubricating gas for the bearings. The friction torque in the MEMS gas bearings is in the 0–100 µN·m scale [17], and the fragile sensor structure may be broken during the system assembling. Therefore, the design of the sensor for the system should take into account the following specifications:

•

•

The sensor should have a relatively high sensitivity to sense the weak force signal. Though a short lever arm can enlarge the force applied to the sensor, the arm installation and alignment between the lever arm and the sensor may turn into a hard connection. Excellent linearity is important. The strain in the sensing region should be less than 1/6–1/5 of the material’s ultimate strain, so as to ensure the measurement linearity, namely less than 500 µm for the silicon [18].


3 of 12


3 of 12

• •

Overload protection is required. Since the silicon-based sensors are fragile, unpredictable 3 of 12the is needed. Thesystem sensor assembling will be connected to the measuring lever arm tocan measure impulsive force in sensor packing, or practical break the target force, and a well-designed form can make the connection more easy and precise. structure, putting the sensor out of work. • A proper sensor form is needed. The sensor will be connected to the lever arm to measure the • A target proper sensor is needed. form The sensor willthe beconnection connectedmore to the lever to measure the force, andform a well-designed can make easy andarm precise. target force, and a well-designed form can make the connection more easy and precise. Micromachines 8, 304form A proper2017, sensor

Figure 1. The schematic diagram of the measurement system for friction torque measurement. Figure the measurement measurementsystem systemfor forfriction frictiontorque torque measurement. Figure1.1.The Theschematic schematicdiagram diagram of of the measurement.

The first two points are mainly determined by the sensing geometry and structural dimensions, The first two points are mainly determined bythe thesensing sensinggeometry geometry and structural dimensions, Thetwo firstare tworelated points are mainly determined bycomponents. structural the latter to the design of sensor Lookingand at the existeddimensions, components thethe latter two are design of components.Looking Lookingatatthe theexisted existed components latter arerelated relatedto tothe the design of sensor sensor components available in two the literature, the probe is often used components. to transmit the force from measurement target to available in the literature, the probe is often used to transmit the force from measurement target available in the literature, the probe is often used to transmit the force from measurement target to to the sensing mechanism and overload protection is implemented by bonding the glass or silicon thethe sensing mechanism and overload protection is implemented by by bonding the glass or silicon wafer sensing mechanism and overload is implemented bonding or wafer silicon wafer to limit the displacement range ofprotection central mesa [11]. Therefore, the probethe andglass boned are to wafer limit the displacement range of central mesamesa [11]. [11]. Therefore, the probe andand boned wafer are also to limit the displacement range of central Therefore, the probe boned wafer are also selected as the elements in the proposed sensor. In the following part of this section, the selected as the elements in the proposed sensor. Insensor. the following part of this section, thesection, investigation also selected thecommon elements in the proposed In the following part of this the be investigation ofas two structures, namely beam structure and membrane structure, will of investigation two common namely beamnamely structure and membrane structure, structure, will be performed, of structures, two common structures, beam structure and membrane will be performed, the design of new sensing structure for the force sensor based on the analysis results will theperformed, design of new sensing for structure the force for sensor basedsensor on thebased analysis results willresults be presented the design of structure new sensing the force on the analysis will be presented and the arrangement of probe and piezoresistors are also conducted. bethe presented and theofarrangement of probe and piezoresistors are also conducted. and arrangement probe and piezoresistors are also conducted. 2.1. Design of Sensor Geometry 2.1. Design 2.1. DesignofofSensor SensorGeometry Geometry InIn thethe design sensing force sensors, thebeam beam structureand andmembrane membrane structure designof sensingstructure structure for structure In the design ofofsensing structure for force force sensors, sensors,the the beamstructure structure and membrane structure areare thethe most commonly used geometry in the published literature. Figure 2 is a schematic view of most commonly used geometryininthe thepublished publishedliterature. literature.Figure Figure22isisaaschematic schematicview view are the most commonly used geometry ofofthe thethe two structures. The are to analyze analyzethe theinfluence influenceononthe thesensor sensor performance two structures. Thetwo twostructures structures are chosen chosen to performance two structures. The two structures are chosen to analyze the influence on the sensor performance from from the geometry. The beam structures, shown in Figure 2a,b consist of four identical beams supporting the geometry. The beam structures, shown in Figure 2a,bconsist consistof of four four identical supporting thefrom geometry. The beam structures, shown in Figure 2a,b identicalbeams beams supporting thethe central mesa, adhered basement. Themembrane membranestructure structure has a similar central mesa,on onwhich whichaaaprobe probeis adhered with with aa has a similar the central mesa, on which probe isis adhered with a basement. basement.The The membrane structure has a similar configuration, but changes into into aa membrane membrane(shown (shownininFigure Figure 2c,d). configuration, butthe thesupporting supportingstructure structure 2c,d). configuration, but the supporting structurechanges changes into a membrane (shown in Figure 2c,d).

Figure2.2.Beam Beamstructure structure (a,b) and force sensor. Figure and membrane membranestructure structure(c,d) (c,d)for forthe the force sensor. Figure 2. Beam structure (a,b) and membrane structure (c,d) for the force sensor.

ToTo analyze commonstructures structuresunder underthe the applied force of 40 mN, analyzethe theperformance performance of of the the two common applied force of 40 mN, To analyze the performance of the two common structures under the applied force of 40 mN, finite element The material materialparameters parametersofofthe the silicon Young’s finite element(FE) (FE)models modelsare areestablished. established. The silicon areare setset as as Young’s finite element (FE) models are established. The parameters of the silicon are set as Young’s 3 material 3 and modulus 2331kg/m kg/m and Poisson’s ratio = 0.278. overall dimension modulusE E= =166 166GPa, GPa,density density ρ = = 2331 Poisson’s ratio ν = ν0.278. The The overall dimension of 3 and Poisson’s ratio ν = 0.278. The overall dimension of modulus E = 166 GPa, density ρ with = 2331 kg/mµm µm × 7000 µm,µm, a 2300 mesa. mesa. In the beam the length, of the thechip chipisis7000 7000 µm × 7000 with a 2300width µm width In thestructure, beam structure, thewidth length, theand chip is 7000 µm × 7000 µm, with 2300 µm200 widthand mesa. In the beam structure, length, width of the are a 1380 µm, in the the membrane, the width thickness and thickness offour the beams four beams are µm, 1380 µm,µm 200 µm50and 50respectively; µm, respectively; in the membrane, and thickness of the four beams are 1380 µm, 200 µm and 50 µm, respectively; in the membrane, the covers allall ofof the space between thethe outer frame and central mesa with thickness of 50 themembrane membrane covers the space between outer frame and central mesa with thickness ofµm. 50 µm. membrane covers all of the space between the outer frame and central mesa with thickness of 50 µm. The force appliedsymmetrically symmetricallyon on the the tip tip of the outer frame is rigidly The force is isapplied of the the probe probeatattwo twopoints pointsand and the outer frame is rigidly The forceFigure is applied symmetrically on theand tip of the probe atalong two points and the outer frame is rigidly fixed. 3 is the strain distribution stress contour the defined path in the beam andand fixed. Figure 3 is the strain distribution and stress contour along the defined path in the beam fixed. Figure 3structures. is the strain distribution and stress contour along structure the defined path inlarger the beam and membrane Under the same force load, the beam acquires strain, membrane Under load, beam acquires largerofstrain, especially structures. at the two ends. The the strainsame comesforce over 500 µε,the which maystructure greatly affect the linearity the especially at the two ends. The strain comes over 500 µε, which may greatly affect the linearity of the

Micromachines Micromachines 2017, 2017, 8, 8, 304 304

of 12 12 44 of

Micromachines 2017, Meanwhile, 8, 304 sensor output.

of 12 the membrane structure only has a strain of about 20 µε, which is4 too membrane structures. Under the same force load, the beam structure acquires larger strain, especially small for the high-sensitivity-needed force sensor. at the two ends. The strain the comes over 500structure µε, whichonly mayhas greatly affect linearity the sensor sensor output. Meanwhile, membrane a strain of the about 20 µε,ofwhich is too output. the membrane structure only has a strain of about 20 µε, which is too small for the small forMeanwhile, the high-sensitivity-needed force sensor. high-sensitivity-needed force sensor.

Figure 3. The strain distribution and stress contour of beam and membrane structures. Figure 3. The strain distribution and stress contour beam and force membrane structures. The stiffness thestrain beam structure and is not enough forof target and the membrane is too Figure 3.ofThe distribution stress contour ofthe beam and membrane structures. rigidThe to sense the force. Thus, a combination of the two structures, namely the beam-membrane stiffness of the beam structure is not enough for the target force and the membrane is(BM) too structure shown 4, can be an effective solution the high sensitivity force sensor. In(BM) stiffness of Figure the Thus, beam is not forfor the target force and the membrane isthe too rigidThe to sense the in force. astructure combination of enough the two structures, namely the beam-membrane newly-designed geometry, four beams are settled atop the membrane, connecting with the central rigid to sense the force. Thus, a combination of the two structures, namely the beam-membrane (BM) structure shown in Figure 4, can be an effective solution for the high sensitivity force sensor. In the mesa along the central lines of the mesa. BM structure utilizes membrane to sensor. enhance structure shown in Figure 4,four can be an effective solution formembrane, the highthe sensitivity force Inthe the newly-designed geometry, beams areThe settled atop the connecting with the central supporting stiffness, while the stress accompanying with the force load will concentrate on the newly-designed geometry, four beams are settled atop the membrane, connecting with the central mesa along the central lines of the mesa. The BM structure utilizes the membrane to enhance the beams, which that sensor still has favorable sensitivity. In the BM to structure, the mesa along the ensures central lines ofstress the mesa. The BMa with structure utilizes the supporting stiffness, while thethe accompanying the force load the willmembrane concentrate onenhance the beams, thickness of the membrane is 30 µm, and the other dimensions are the same to the abovementioned supporting stiffness, while thestill stress with theInforce load will concentrate on the which ensures that the sensor has accompanying a favorable sensitivity. the BM structure, the thickness of membrane structure; for the size of beams, thickness is also the only changed parameter that turns beams, which ensures still has a are favorable sensitivity. In the BM structure, the the membrane is 30 µm,that andthe the sensor other dimensions the same to the abovementioned membrane into 20 µm. To compare structures, FE model of BMare structure is also established the structure; the size ofthe beams, is the also the only changed parameter turns intoand 20 µm. thickness offor the membrane isthree 30 thickness µm, and the other dimensions the same tothat the abovementioned simulated relationships between the maximum von stressisinalso theestablished beam/membrane and applied To compare the three for structures, FE model of BMMises structure and the simulated membrane structure; the size the of beams, thickness is also the only changed parameter that turns force are shown in Figure 5.maximum It can be found that all the structures have a good linearity relationship relationships von Mises inof the beam/membrane and applied force into 20 µm. To between compare the the three structures, the FEstress model BM structure is also established and the between theinforce and5.the stress, and the BM structure has a more favorable stress are shown Figure It accompanying can be that all the structures have good linearity relationship simulated relationships between thefound maximum von Mises stress in the abeam/membrane and applied that maintains measurement sensitivity stress, and keeps the totalstructure strain below 500 µε to avoidstress possible between the force the 5. accompanying hasaagood more favorable that force are shown inand Figure It can be found thatand all the BM structures have linearity relationship nonlinear deformation insensitivity the sensing Thetotal strain distribution ofµε thetoBM structure is nonlinear shown in maintains measurement andregion. keeps the below 500 possible between the force and the accompanying stress, and strain the BM structure hasavoid a more favorable stress Figure 6, andinthe total strain of thedistribution beam in is about 275 µε, conforming to the linearity deformation themaximum sensing region. The strain of the BMbelow structure in Figure 6, that maintains measurement sensitivity and keeps the total strain 500 isµεshown to avoid possible requirement. and the maximum totalinstrain of the beam in isThe about 275distribution µε, conforming to the nonlinear deformation the sensing region. strain of the BMlinearity structurerequirement. is shown in Figure 6, and the maximum total strain of the beam in is about 275 µε, conforming to the linearity requirement.

The designed designed beam-membrane structure. (a) 3D view; (b) cross-sectional view. Figure 4. The

Figure 4. The designed beam-membrane structure. (a) 3D view; (b) cross-sectional view.

Micromachines 2017, 8, 304 Micromachines 8, 304 Micromachines 2017,2017, 8, 304

5 of 12 5 of 125 of 12

Figure 5. Sensitivity comparison of three structures. Figure 5. Sensitivity comparison of three structures. Figure 5. Sensitivity comparison of three structures.

Figure 6. The strain distribution and stress contour of beam-membrane structure.

Figure 6. The strain distribution and stress contour of beam-membrane structure. Figure 6. The strain distribution and stress contour of beam-membrane structure. 2.2. Design of Sensor Components

2.2. Design of Sensor Components Asofthe signalComponents transforming component in the sensor, piezoresistors play an important role in the 2.2. Design Sensor As the signal transforming component in the piezoresistors play an important role force measurement. For a piezoresistor oriented insensor, [110] direction, the variation of the resistance canin the As the signal transforming component in the sensor, piezoresistors play an important role in the expressed as ∆RFor [19]. forcebemeasurement. a piezoresistor oriented in [110] direction, the variation of the resistance can force measurement. For a piezoresistor oriented ∆R 1 in [110] direction, the variation of the resistance can be expressed as ΔR [19]. = π44 (σl − σt ) (1) be expressed as ΔR [19]. R 2 In Equation (1), R, ∆R are the original and variation resistance of the piezoresistor, σ1 , σt are ΔR 1 directions of the piezoresistor, π is the shearing the stresses in the longitudinal and transverse 44 (1) ΔR R = 12 π 44 σ l − σ t piezoresistance coefficient. The piezoresistors arrangement is shown in Figure 7, for R1 and R3 the (1) = π 44 σ l − σ t R stress 2 and they will obtain a positive change; meanwhile, beam normal stress are the longitudinal

( (

) )

In Equation (1), R, ΔR are the original and variation resistance of the piezoresistor, σ1, σt are the In Equation (1), R, ΔR are and the original and directions variation resistance of the piezoresistor, σ1, σshearing t are the stresses in the longitudinal transverse of the piezoresistor, π44 is the stresses in the longitudinal and transverse directions of the piezoresistor, π 44 is the shearing piezoresistance coefficient. The piezoresistors arrangement is shown in Figure 7, for R1 and R3 the piezoresistance coefficient. piezoresistors is obtain shown ainpositive Figure 7, for R1 meanwhile, and R3 the beam normal stress are theThe longitudinal stressarrangement and they will change; beam normal stress are the longitudinal stress and they will obtain a positive change; meanwhile,

Micromachines 2017, 8, 304 Micromachines 2017, 8, 304

6 of 12 6 of 12

the normal stress turns into transverse stress for R2 and R4, and their variation will be negative. normal stress turns into transverse stress for R2 and R4 , and their variation will be negative. Therefore, athe full Wheatstone bridge can be constructed by the four piezoresistors, as Figure 7 shows. Therefore, a full Wheatstone bridge can be constructed by the four piezoresistors, as Figure 7 shows.

Figure 7. The piezoresistors arrangement in the beam-membrane (BM) structure sensor.

Figure 7. The piezoresistors arrangement in the beam-membrane (BM) structure sensor. Another component is the sensor probe. The probe should be inflexible and lightweight to eliminate the influence on the sensor performances from the incorporation of probe. Therefore, a tubular Another component is the sensor probe. The probe should be inflexible and lightweight to shape, stainless-steel-made probe is chosen. Moreover, a sleeve is used to enhance the installation eliminate the influence the sensor performances from incorporation probe. Therefore, a accuracy, which on is marked red in Figure 4. According to the FEthe analysis, the maximumof displacement of thestainless-steel-made central mesa is 3 µm when the 40 mNisforce is applied. Therefore, toaprotect the sensing structure tubular shape, probe chosen. Moreover, sleeve is used to enhance the from broken, the Pyrex glass wafer is attached on the bottom of the chip with a gap of 3 µm from the installation accuracy, which is marked red in Figure 4. According to the FE analysis, the maximum bottom of the central mesa.

displacement of the central mesa is 3 µm when the 40 mN force is applied. Therefore, to protect the 3. Sensor Realization sensing structure from broken, the Pyrex glass wafer is attached on the bottom of the chip with a gap of 3 µm from bottom of the central mesa. 3.1.the Fabrication The main fabrication process of the sensor chip is shown in Figure 8. A double-side polished,

3. Sensor Realization N-type (100)-oriented silicon wafer was used as the start material and a total of eight masks were needed. The detailed fabrication process is as follows:

3.1. Fabrication (1)

The fabrication began with the inductively coupled plasma (ICP) etching at the back side to form the working gap between the bottom of the central mesa and bonded glass (Figure 8a); The main of the sensor chip shown in Figure A double-side (2) fabrication Then the boronprocess ion diffusion and driven-in processiswere conducted after the8. piezoresistors were polished, patternedsilicon on the front side was (Figure 8b); as the start material and a total of eight masks were N-type (100)-oriented wafer used (3) The interconnections of piezoresistors were realized by Al sputtering and the electrodes were needed. The detailed fabrication process is as follows: simultaneously formed (Figure 8c); (4) The deposited SiOinductively the protection layer for the following KOH (1) The fabrication beganSiwith the coupled plasma (ICP) etching at etchant the back side to 3 N4 and 2 were utilized as based anisotropic wet etching, and the central mesa was shaped by wet etching from the back form the working gap between the bottom of the central mesa and bonded glass (Figure 8a); side (Figure 8d); (2) Then the boron ion diffusion and driven-in process were conducted after the piezoresistors (5) The beam-membrane structure was formed by ICP etching on the front side (Figure 8e); were patterned onglass thewafer frontwas side (Figure 8b); (6) A Pyrex bonded onto the bottom of silicon wafer by anodic bonding (Figure 8f). A Cr/Au bi-layer was deposited on the glassrealized wafer to avoid electrostatic adhesion the (3) The interconnections of piezoresistors were by Al sputtering andbetween the electrodes were mesa and glass during the bonding or overloaded process.

simultaneously formed (Figure 8c); The finished sensor chip is shown in Figure 9. (4) The deposited Si3N4 and SiO2 were utilized as the protection layer for the following KOH etchant based anisotropic wet etching, and the central mesa was shaped by wet etching from the back side (Figure 8d); (5) The beam-membrane structure was formed by ICP etching on the front side (Figure 8e); (6) A Pyrex glass wafer was bonded onto the bottom of silicon wafer by anodic bonding (Figure 8f). A Cr/Au bi-layer was deposited on the glass wafer to avoid electrostatic adhesion between the mesa and glass during the bonding or overloaded process.

Micromachines Micromachines 2017, 2017, 8, 8, 304 304 Micromachines 2017, 8, 304

77 of of 12 12 7 of 12

Figure 8. 8. Main fabrication process of the sensor chip. (a) Etching at the back back side side to to form form the the gap gap Figure Main fabrication fabrication process process of of the the sensor sensor chip. chip. (a) (a) Etching Etching at at the the Figure 8. Main back side to form the gap between the the bottom bottom of of the the central central mesa mesa and and bonded bonded glass; glass; (b) (b) Boron Boron ion ion diffusion diffusion and and driven-in driven-in between between the bottom of the central mesa and bonded glass; (b) Boron ion diffusion and driven-in process for piezoresistors on the front side; (c) Al sputtering to form the interconnections of process for for piezoresistors piezoresistors on on the the front front side; side; (c) (c) Al Al sputtering sputtering to to form form the the interconnections interconnections of of process piezoresistors and the electrodes; (d) The central mesa was shaped by wet etching from the back side; piezoresistors and and the the electrodes; electrodes; (d) (d) The The central central mesa mesa was was shaped shaped by by wet wet etching etchingfrom fromthe theback backside; side; piezoresistors (e) The The beam-membrane beam-membrane structure structure was was formed formed by by inductive inductive coupled coupled plasma plasma (ICP) (ICP) etching etching on on the the (e) (e) The beam-membrane structure was formed by inductive coupled plasma (ICP) etching on the front side; side; (f) (f) Pyrex Pyrex glass glass wafer wafer was was bonded bonded onto onto the the bottom of of siliconwafer wafer byanodic anodic bonding. front front side; (f) Pyrex glass wafer was bonded onto the bottom bottom of silicon silicon wafer by by anodic bonding. bonding.

Figure The fabricated fabricatedsensor sensorchip. chip.(a) (a)Photo Photoofofwhole whole chip SEM piezoresistors and wire Figure 9. 9. The chip (b)(b) SEM of of piezoresistors and Al Al wire (c) Figure 9.ofThe fabricated sensor chip. (a) Photo ofbase whole chip (b) SEM of piezoresistors and Al wire (c) (c) SEM the beam, membrane and island (the of probe). SEM of the beam, membrane and island (the base of probe). SEM of the beam, membrane and island (the base of probe).

3.2. Sensor Packing 3.2. Sensor Packing To theprecision precisionofofthe thepackaging, packaging, a tailor-made organic glass fixture is designed. To guarantee the a tailor-made organic glass fixture is designed. The To guarantee the precision of the packaging, a tailor-made organic glass fixture is designed. The The fixture has two grooves for the positioning of the sensor and printed circuit board (PCB), and a hole fixture has two grooves for the positioning of the sensor and printed circuit board (PCB), and fixture has two grooves for the positioning of the sensor and printed circuit board (PCB), and a hole for the probe. The packaging process is illustrated in Figure 10. Firstly, the sensor chip and PCB were for the probe. The packaging process is illustrated in Figure 10. Firstly, the sensor chip and PCB were aligned with the grooves in the glass fixture and bonded by adhesive; after that, the probe was put aligned with the grooves in the glass fixture and bonded by adhesive; after that, the probe was put

Micromachines 2017, 8, 304 Micromachines 2017, 8, 304 Micromachines 2017, 8, 304

8 of 12 8 of 12 8 of 12

aligned withand the grooves the the glass fixture and of bonded by adhesive; after the that,electrical the probe was put into the hole adhered in onto central mesa the sensor chip; finally, connection into the hole and adhered onto the central mesa of the sensor chip; finally, the electrical connection into the hole and adhered onto the central mesa of the sensor chip; finally, the electrical connection between the sensor and instruments was conducted with golden leads by wire bonder after the between sensor and was with golden golden leads leads by by the thewire wirebonder bonderafter afterthe the betweenthe the sensor and instruments instruments was conducted conducted with necessary thermal solidification of adhesive. necessary thermal solidification of adhesive. necessary thermal solidification of adhesive.

Figure 10. Packaging of the force sensor. Figure 10. Packaging Packaging of the force Figure force sensor. sensor.

4. Sensor Testing 4.4.Sensor SensorTesting Testing 4.1. Experimental Setup 4.1.Experimental ExperimentalSetup Setup 4.1. The calibration system, shown in Figure 11, uses an analytical balance and high-precision The calibration calibration system, system, shown shown in in Figure 11, uses an analytical balance and high-precision The 11, force uses for an the analytical balance high-precision positioning stage to generate and displayFigure the input sensor; the force and can be obtained by positioning stage to generate and display the input force for the sensor; the force can be obtained by positioning stage to generate and display the input force for the sensor; the force can be obtained by 2 multiplying the measured weight and gravity acceleration (10 m/s ). The positioning stage consists 2 2 multiplying the the measured weight and gravity acceleration (10(10 m/s ). ). The positioning stage consists of multiplying measured weight and gravity acceleration m/s The positioning stage consists of a tri-dimensional positioning stage and a piezoelectric ceramics with the resolution up to nm a tri-dimensional positioning stage andand a piezoelectric ceramics with with the resolution up to nm scale. of a tri-dimensional positioning stage a piezoelectric ceramics the resolution up sensor to nm scale. The tri-dimensional positioning stage adjusts the sensor location, making the tip of the The tri-dimensional positioning stage adjusts the sensor location, making the tip of the sensor probe scale. The tri-dimensional positioning stage adjusts the sensor location, making the tip of the sensor probe very close to the sensing plate of balance; then, the piezoelectric ceramics moves the sensor very close to the sensing plate ofplate balance; then, thethen, piezoelectric ceramics moves themoves sensorthe with nm probe very close to the sensing balance; sensor with nm displacement, pressuring theofsensor probe tothe thepiezoelectric balance. Theceramics excitation voltage for the displacement, pressuring the sensor probe to the balance. The excitation voltage for the Wheatstone with nm displacement, pressuring the sensor probe to the balance. The excitation voltage for the Wheatstone bridge in the sensor is provided by the power source (ITECH IT6322, ITECH Electronic bridge in the sensor isthe provided by the power source (ITECH IT6322, ITECH Electronic Co., Ltd., Wheatstone bridge in sensor is provided by the power source (ITECH IT6322, ITECH Electronic Co., Ltd., Nanjing, China), and the output voltage can be measured by a digital multimeter (FLUKE Nanjing, China), and the output voltage be measured by a digital (FLUKE (FLUKE 8845A, Co., Ltd., Nanjing, China), and the outputcan voltage can be measured by amultimeter digital multimeter 8845A, Fluke Inc., Everett, WA, USA). Fluke Inc., Everett, WA, USA). 8845A, Fluke Inc., Everett, WA, USA).

Figure 11. The setup for sensor Figure sensor calibration. calibration. Figure 11. The setup for sensor calibration.

4.2. Results and Discussion 4.2. Results and Discussion The static characteristics of the force sensor were tested by the abovementioned calibration TheThe static characteristics sensor were tested system. excitation voltageofforthe theforce Wheatstone bridge was by 5 Vthe DCabovementioned and the applied calibration force rose system. The excitation voltage for the Wheatstone bridge was 5 V DC and the applied force rose


9 of 12

4.2. Results and Discussion The static2017, characteristics of the force sensor were tested by the abovementioned calibration9 system. Micromachines 8, 304 of 12 The excitation voltage for the Wheatstone bridge was 5 V DC and the applied force rose from 0 to about 0 to decreased about 50 mN 0 in one test. The results experimental results were shown Figurethe 50 from mN and toand 0 indecreased one test. to The experimental were shown in Figure 12.inWhen 12. When the applied force is larger than 40 mN, the increase tendency of output voltage became applied force is larger than 40 mN, the increase tendency of output voltage became tardy and the tardy and the at protection glass thesensor bottomchip of the sensor chip began toIn play role. In the designed protection glass the bottom ofatthe began to play its role. theitsdesigned linear range of linear range of 0–40 mN, the results were fitted with the least square method and the relationship 0–40 mN, the results were fitted with the least square method and the relationship between the applied between the applied force F (mN) and output voltage V (mV) can be expressed as Equation (2) as force F (mN) and output voltage V (mV) can be expressed as Equation (2) as follows: follows:

0.127F 4.805 VV ==0.127 F+ + 4.805

(2) (2)

Obviously, the measured sensitivity was 0.127 mV/mN, and the maximum non-linearity was Obviously, the measured sensitivity was 0.127 mV/mN, and the maximum non-linearity was 1.97 %FS and the total measurement accuracy is 2.13 %FS. The main goal of a measurement range of 1.97 %FS and the total measurement accuracy is 2.13 %FS. The main goal of a measurement range of 40 mN and a sensitivity of about 0.1 mV/mN are achieved. The other tested parameters of the sensor 40 mN and a sensitivity of about 0.1 mV/mN are achieved. The other tested parameters of the sensor are shown in Table 1.

are shown in Table 1.

Figure 12. The results of sensor calibration and overload protection. Figure 12. The results of sensor calibration and overload protection.

Table 1. Performance of the beam-membrane (BM) micro-force sensor. Table 1. Performance of the beam-membrane (BM) micro-force sensor.

Parameter Parameter Zero point Zero pointoffset offset(mV) (mV) Full Fullscale scalespan span(mV) (mV) Sensitivity Sensitivity(mV/mN) (mV/mN) Nonlinearity(%FS) (%FS) Nonlinearity TemperatureDrift DriftFactor Factor (/°C) (/◦ C) Temperature Time Drift (%FS/h) Time Drift (%FS/h) Hysteresis (%FS) Hysteresis (%FS) Repeatability (%FS) Repeatability (%FS) Accuracy (%FS) Accuracy (%FS)

Value Value 4.805 4.805 9.885 9.885 0.127 0.127 1.97 1.97 −0.00628 −0.00628 0.30473 0.30473 0.548 0.548 0.761 0.761 2.13 2.13

5. Sensor Application

5. Sensor Application

5.1. Experimental Setup

5.1. Experimental Setup

The actual measurement system is sketched in Figure 13. The ultra-precision computer-controlled The actual measurement system is sketched in Figure 13. The ultra-precision rotary table generates the original rotation by its step motor, which is transmitted to the shaft of the computer-controlled rotary table generates the original rotation by its step motor, which is gastransmitted bearings by andbearings a silicon-based lever arm. and Thealever arm is fabricated laser to the the force shaft sensor of the gas by the force sensor silicon-based lever arm.by The lever arm is fabricated by laser scribing apparatus, and the final size is 50 mm long, 1 mm wide and


10 of 12


10 of 12

304 12 400Micromachines µm thick. 2017, The 8,sensor probe is horizontal and vertically connected to the lever arm which10isoffixed onto the shaft of MEMS bearing.

400 µm thick. The sensor probe is horizontal and vertically connected to the lever arm which is fixed

scribing apparatus, and the final size is 50 mm long, 1 mm wide and 400 µm thick. The sensor probe is onto the shaft of MEMS bearing. horizontal and vertically connected to the lever arm which is fixed onto the shaft of MEMS bearing.

Figure 13. The system for friction torque measurement in micro-electro-mechanical system (MEMS) Figurebearing. 13. The system for friction torquetorque measurement in micro-electro-mechanical system Figure 13. The system for friction measurement in micro-electro-mechanical system(MEMS) (MEMS) bearing. bearing.

The MEMS bearing is installed in in thethe central hole of of thethe rotary table by by a stationary platform with The MEMS bearing is installed central hole rotary table a stationary platform The MEMS bearing is installed in the central hole of the rotary table by a stationary platform necessary gas passage. The force sensor is adhered on an off-center settled fixture with an adjustable with necessary gas passage. The force sensor is adhered on an off-center settled fixture with an withfrom necessary gas passage. The whole force sensor is adhered an off-center fixture with an distance the table center. The platform isplatform fixedononto a horizontal stand. The assembled adjustable distance from the table center. The whole is fixed ontosettled a horizontal stand. The adjustable distance from the table center. The whole platform is fixed onto a horizontal stand. The system is shown in Figure 14. in High purity is led to the gas bearings and assembled system is shown Figure 14. nitrogen High purity nitrogen is led to the for gassuspension bearings for assembled system is shown in Figure 14. High purity nitrogen is led to the gas bearings for lubrication. pressure for gas bearing is 4.5PSI, while 4.86PSI forwhile the lower gasfor bearing suspensionThe andgas lubrication. Theupper gas pressure for upper gas bearing is 4.5PSI, 4.86PSI the suspension and lubrication. The gas pressure for upper gas bearing is 4.5PSI, while 4.86PSI for the to lower balance the weighttoofbalance the lever shaft andlever rotor.arm, Theshaft length forceThe armlength for theoflever gas bearing thearm, weight of the andofrotor. forceand armsensor for lower gas bearing to balance the weight of the lever arm, shaft and rotor. The length of force arm for is set 5 mm be changed by moving connection point between the sensor probe and the as lever andwhich sensorcan is set as 5 mm which can bethe changed by moving the connection point between the lever and sensor is set as 5 mm which can be changed by moving the connection point between thethe sensor probe and lever lever arm. sensor probe and leverarm. arm. Running scheme of rotary table programmed torotate rotatefrom fromstillness stillnesstoto touniform uniform Running scheme of rotary table isisis programmed stillness uniform rotation at Running schemethe ofthe the rotary table programmedto to rotate from rotation ◦ /s. the speed of 1°/s. The output voltage of the force sensor was recorded by an oscillograph and theatspeed of 1 The output voltage of the force sensor was recorded by an oscillograph and shown at the speed of 1°/s. The output voltage of the force sensor was recorded by an oscillograph and shown in Figure 14b. The maximum static friction torque occurs when the maximum voltage shown in The Figure 14b. Thestatic maximum static friction torque occurs when the maximum voltage during is is in Figure 14b. maximum friction torque occurs when the maximum voltage is output output during the starting process. output during the starting process. the starting process.

(a)

(a)

(b)

(b)

Figure 14. Photo of the measurement platform (a) and the assembled system (b).

Figure14. 14.Photo Photoofofthe themeasurement measurement platform platform (a) (b). Figure (a)and andthe theassembled assembledsystem system (b). 5.2. Results and Discussion

5.2.5.2. Results and Results andDiscussion Discussion

As shown in Figure 15, the maximum voltage is obtained at the end of the pulse-on process.

As shown 15, voltage is obtained obtained the endthe the pulse-on process. As shownininFigure Figure 15,the themaximum maximum voltagetorque is atatthe end ofofthe pulse-on process. Correspondingly, the maximum starting friction is achieved with maximum starting Correspondingly, the maximum starting friction torque is achieved with the maximum starting friction Correspondingly, the time. maximum friction torque achieved with maximum starting friction force at this Three starting representative curves areiscollected with the the maximum voltage of force at this time. Three representative curves are collected with the maximum voltage of 6.856 mV, 6.856 force mV, 6.845 mVtime. and 6.862 mV, respectively. The average value of the maximum voltagesvoltage is 6.854 of friction at this Three representative curves are collected with the maximum 6.845 mV and 6.862 The average of the maximum voltagesvoltages is 6.854ismV and 6.856 mV, 6.845 mVmV, andrespectively. 6.862 mV, respectively. Thevalue average value of the maximum 6.854 the matching force is 16.314 mN, which is calculated according to the calibration results in Figure 14. Hence, the maximum starting friction torque is 80.669 µN·m, considering the length of the force arm is 5 mm.


11 of 12

mV and the matching force is 16.314 mN, which is calculated according to the calibration results in Micromachines 8, 304the maximum starting friction torque is 80.669 µN·m, considering the length of the 11 of 12 Figure 14.2017, Hence,

force arm is 5 mm. Passing the pulse-on process, the bearings began to set up uniform rotation and the output of Passing the pulse-on process, the bearings began set upvoltage uniform rotation thethree output the force sensor maintained in a stable-state range. The to average is 5.559 mV and for the of the force sensor maintained in a stable-state range. The average voltage is 5.559 mV for representative curves in the steady stage. The matching friction force is 5.937 mN and the the three representative curves in the steady stage. The matching friction force is 5.937 mN and the corresponding torque is 29.685 µN·m.

corresponding torque is 29.685 µN·m.

Figure 15. The output voltage of the sensor under start-up and steady stage. Figure 15. The output voltage of the sensor under start-up and steady stage.

As observed from the figure, the three curves show high repeatability, only with small fluctuations observed from the figure, three curves show high repeatability, with small at steadyAsstage. Possible reasons for thethe fluctuation can be unstable gas flowing inonly the bearing or gas fluctuations at engine steady stage. turbulence in the [20]. Possible reasons for the fluctuation can be unstable gas flowing in the bearing or gas turbulence engine [20]. that the measurement range and sensitivity of the sensor With the above results, itinisthe demonstrated With the above results, it is demonstrated that the range and sensitivity of theand we designed and fabricated can meet the requirements for measurement measuring large starting friction torque sensor we designed and fabricated can meet the requirements for measuring large starting friction kinetic friction torque in rotating MEMS machines. This sensor, which incorporates a membrane into torque and kinetic friction torque in rotating MEMS machines. This sensor, which incorporates a conventional four-beam structure, has the capability to eliminate the problem of small measurement membrane into conventional four-beam structure, has the capability to eliminate the problem of range with neat beam structure and low sensitivity with neat membrane structure. small measurement range with neat beam structure and low sensitivity with neat membrane structure. 6. Conclusions

force sensor with beam-membrane structure has been designed, fabricated and tested for use in 6.AConclusions a friction torque measurement system. The proposed sensing structure combines the conventional A force sensor with beam-membrane structure has been designed, fabricated and tested for use beam structure and membrane structures, which gives the sensor a relatively high sensitivity and in a friction torque measurement system. The proposed sensing structure combines the conventional highbeam reliability. A hollow stainlessstructures, steel probewhich is bonded fabricated sensor chip with the and help of structure and membrane gives to thethe sensor a relatively high sensitivity a tailor-made glassAfixture. According toprobe the results of sensor the sensitivity, maximum high reliability. hollow stainless steel is bonded to the calibration, fabricated sensor chip with the help non-linearity error and measurement accuracy of the developed sensor are 0.127 mV/mN, 1.97 %FS of a tailor-made glass fixture. According to the results of sensor calibration, the sensitivity, andmaximum 2.13 %FS, non-linearity respectively. error The bonded glass wafer at the bottom of the sensor chip actmV/mN, as overload and measurement accuracy of the developed sensor arecan 0.127 protection when the applied force is beyond the designed range. With the aim ofsensor evaluating theact usage 1.97 %FS and 2.13 %FS, respectively. The bonded glass wafer at the bottom of the chip can as force overload protection when for the friction appliedtorque force is beyond the designed range. With the aim of of the sensor in the system measurement, practical experiments for estimation evaluating theinusage of bearings the force sensor in the system for friction torque measurement, practical of friction state the gas are conducted. The results show promising variation tendency experiments for estimation of friction state in the gas bearings are conducted. The results showinto of friction in the gas bearing, demonstrating the feasibility of embedding a MEMS force sensor variation of friction in the torque gas bearing, demonstrating theµN feasibility of the promising measurement systemtendency to investigate the friction down to the scale of 100 ·m. Therefore, embedding a MEMS force sensor into the measurement system to investigate the friction torque the sensor has great potential application in in-situ measuring micro friction force and torque under down to the scale of 100 µN·m. Therefore, the sensor has great potential application in in-situ pulse-on stage and stable stage in rotating MEMS machines.

measuring micro friction force and torque under pulse-on stage and stable stage in rotating MEMS machines. Acknowledgments: The work was supported by the Young Scientists Fund of the National Natural Science Foundation of China (Grant Nos. 51605085, 51703180) and the Postdoctoral Science Foundation of China (Grant No. 2016M590229). Author Contributions: Huan Liu and Zhongliang Yu conceived and designed the experiments; Yan Liu and Xudong Fang completed the modeling and simulation; Zhongliang Yu fabricated the chip; Huan Liu performed the experiments and wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.


12 of 12

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19.

20.

Alan, H.E. Introduction to power MEMS. In Multi-Wafer Rotating MEMS Machines; Jeffrey, H.L., Ed.; Springer: New York, NY, USA, 2002; pp. 1–8. Kim, D.; Lee, S.; Jin, Y.; Desta, Y.; Bryant, M.D.; Goettert, J. Micro gas bearings fabricated by deep X-ray lithography. Microsyst. Technol. 2004, 10, 456–461. [CrossRef] Spakovszky, Z.S.; Liu, L.X. Scaling laws for ultra-short hydrostatic gas journal bearings. J. Vib. Acoust. 2005, 127, 254–261. [CrossRef] Lee, Y.-B.; Kwak, H.-D.; Kim, C.-H.; Lee, N.-S. Numerical prediction of slip flow effect on gas-lubricated journal bearings for MEMS/MST-based micro-rotating machinery. Tribol. Int. 2005, 38, 89–96. [CrossRef] Ling, R.; Ke-Qin, Z.; Xiao-Li, W. Effects of the slip velocity boundary condition on the characteristics of microbearings. J. Micromech. Microeng. 2004, 14, 116–124. Daejong, K.; Sanghoon, L.; Michael, D.B.; Frederick, F.L. Hydrodynamic performance of gas microbearings. J. Tribol. 2004, 126, 711–718. Ehrich, F.F.; Jacobsonm, S.A. Development of high-speed gas bearings for high-power density microdevices. J. Eng. Gas Turbines Power 2002, 125, 141–148. [CrossRef] Savoulides, N.; Breuer, K.S.; Jacobson, S.; Ehrich, F.F. Low-order models for very short hybrid gas bearings. J. Tribol. 2000, 123, 368–375. [CrossRef] Bouyer, J.; Fillon, M. Experimental measurement of the friction torque on hydrodynamic plain journal bearings during start-up. Tribol. Int. 2011, 44, 772–781. [CrossRef] McCarthy, M.; Waits, C.M.; Ghodssi, R. Dynamic friction and wear in a planar-contact encapsulated microball bearing using an integrated microturbine. J. Microelectromech. Syst. 2009, 18, 263–273. [CrossRef] Weizhong, W.; Yulong, Z.; Yafei, Q. A novel integrated multifunction micro-sensor for three-dimensional micro-force measurements. Sensors 2012, 12, 4051–4064. Zhai, J.; Karuppasamy, K.; Zvavanjanja, R.; Fisher, M.; Fisher, A.C.; Gould, D.; How, T. A sensor for needle puncture force measurement during interventional radiological procedures. Med. Eng. Phys. 2013, 35, 350–356. [CrossRef] [PubMed] Bian, T.; Yulong, Z.; Zhuangde, J. The novel structural design for pressure sensors. Sens. Rev. 2010, 30, 305–313. Jordan, A.; Büttgenbach, S. Micromechanical force sensors based on su-8 resist. Microsyst. Technol. 2012, 18, 1095–1101. [CrossRef] Palli, G.; Pirozzi, S. A miniaturized optical force sensor for tendon-driven mechatronic systems: Design and experimental evaluation. Mechatronics 2012, 22, 1097–1111. [CrossRef] Yoshikawa, G.; Akiyama, T.; Gautsch, S.; Vettiger, P.; Rohrer, H. Nanomechanical membrane-type surface stress sensor. Nano Lett. 2011, 11, 1044–1048. [CrossRef] [PubMed] Zoltán, S.S. High-speed gas bearings for microturbomahcinery. In Multi-Wafer Rotating MEMS Machines; Jeffrey, H.L., Ed.; Springer: New York, NY, USA, 2002; pp. 191–279. Yan, L.; Yulong, Z.; Sun, L. An improved structural design for accelerometers based on cantilever beam-mass structure. Sens. Rev. 2012, 32, 222–229. [CrossRef] Tian, B.; Zhao, Y.; Jiang, Z.; Zhang, L.; Liao, N.; Liu, Y.; Meng, C. Fabrication and structural design of micro pressure sensors for tire pressure measurement systems (TPMS). Sensors 2009, 9, 1382–1393. [CrossRef] [PubMed] Beerschwinger, U.; Reuben, R.L.; Yang, S.J. Friction study of micromotor bearings. Sens. Actuators A Phys. 1997, 63, 229–241. [CrossRef] © 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/).