Flexible Tactile Sensor Array Based on Aligned MWNTs-PU ... - MDPI

micromachines Article

Flexible Tactile Sensor Array Based on Aligned MWNTs-PU Composited Sub-Microfibers Weiting Liu 1 , Xiaoying Cheng 1,2, *, Xiaodong Ruan 1 and Xin Fu 1 1 2

*

State Key Laboratory of Fluid Power & Mechatronic Systems, Zhejiang University, No. 38 Zheda Road, Hangzhou 310027, China; [email protected] (W.L.); [email protected] (X.R.); [email protected] (X.F.) Faculty of Mechanical Engineering and Automation, Zhejiang Sci-Tech University, No. 928 Second Avenue, Hangzhou 310018, China Correspondence: [email protected]; Tel.: +86-571-8795-3395

Received: 27 February 2018; Accepted: 20 April 2018; Published: 24 April 2018







Abstract: This present paper describes a novel method to fabricate tactile sensor arrays by producing aligned multi-walled carbon nanotubes (MWNTs)-polyurethane (PU) composite sub-microfiber (SMF) arrays with the electrospinning technique. The proposed sensor was designed to be used as the artificial skin for a tactile sensation system. Although thin fibers in micro- and nanoscale have many good mechanical characteristics and could enhance the alignment of MWNTs inside, the high impedance as a consequence of a small section handicaps its application. In this paper, unidirectional composite SMFs were fabricated orthogonally to the parallel electrodes through a low-cost method to serve as sensitive elements (SEs), and the impedances of SEs were measured to investigate the changes with deformation caused by applied force. The particular piezoresistive mechanism of MWNTs disturbed in SMF was analyzed. The static and dynamic test results of the fabricated tactile sensor were also presented to validate the performance of the proposed design. Keywords: flexible tactile sensors; sub-microfiber array; nanocomposite; carbon nanotubes; electrospinning

1. Introduction To simulate the human skin, which has four types of mechanoreceptors to archive high spatial definition and keep flexibility and compliance simultaneously, extensive research has been conducted on developing different kinds of transduction methods for tactile sensors that can be embedded in the structure of a robot or prosthetic hand [1–4]. Among these mechanisms, the piezoresistive method is prevalent for its simple and low-cost preparation process with good flexibility [5]. Moreover, piezoresistive composite materials made up of the polymeric matrix and conductive fillers have been widely studied over the last decades because of their higher sensitivity and higher compatibility with complex curves than rigid materials such as microelectromechanical systems (MEMS) piezoresistors [6–8]. Although the composites are intrinsically sensitive and flexible, the bulky size (especially in the thick direction) of these materials would impede the full utilization of these good properties [9]. In contrast, the form of fine fiber can improve the sensitivity and flexibility of materials [10,11], and unlike direct drawing, template synthesis, self-assembly and phase separation techniques, electrospinning is a more effective and convenient method able to fabricate polymer-based composite fibers in various diameters [12]. However, the nano- and microfibers of functional composites are well applied in many areas, such as biological and chemical sensors [13], whereas the reports about the tactile sensing are few [14,15]. Therefore, to develop a highly sensitive, simply structured and low-cost tactile sensing device, this paper presents a novel proposal of a flexible tactile sensor based on sensitive composited SMF arrays fabricated by a modified electrospinning method. The composite here is a dielectric polymer Micromachines 2018, 9, 201; doi:10.3390/mi9050201

www.mdpi.com/journal/micromachines

Micromachines 2018, 9, 201

2 of 13

with conductive particles filled. The electrical resistivity of this material would drop continuously when the volume fraction of dopant increases until the percolation threshold is reached. At this point, the resistance of the composite changes dramatically along with the variation of external force applied on it [16]. Although the associated theory in bulk size has already been extensively studied, the counterpart in slim structure has not been reported other than morphological analysis [17,18]. In this work, conductive polymer composite is prepared in aligned thin fibers of sub-micrometer scale to archive highly orientated arrangement of conductive fillers. Unidirectional composite sub-microfibers (SMFs) are orthogonal to the parallel electrodes to form resistance networks to decrease the high impedance of one single composite fiber. Then two sets of n SEs parallel arrays are perpendicular to each other to form a matrix so that 2n sensitive elements (SEs) can make both electrical connection and electronic interface much easier than N × n SEs. Meanwhile, SMFs are developed by electrospinning process modified by rotating technique which has a much simpler processing than other solutions of producing the slim structure. In the previous study, it was found that a polymer integrated with carbon nanotubes could achieve good piezoresistive performance because of the property of the latter material [19]. In this paper, multi-walled carbon nanotubes (MWNTs) are distributed into polyurethane (PU) to realize the tactile sensor. By a modified electrospinning technique, composite fibers are fabricated onto a flexible printed circuit board (FPCB), which is demonstrated later. The resistivity-force characteristics of these sensors are investigated. Other relevant data such as static and dynamic properties are also shown. The mechanism of the sensing process and the structure of the proposed sensor is described in Section 2. Section 3 is about the fabrication of sensing elements through the modified electrospinning method. Then, the test results are shown and discussed in Section 4. 2. Sensor Mechanism and Design 2.1. Mechanism Analysis In this work, sensitive elements are made of PU with doped MWNTs. Carbon particles are utilized as one of the most common conductive filling materials for their high electrical conductivity and stability. When approaching percolation threshold, MWNTs can not only improve the sensitive of piezoresistive composite much more than other allotropes but also decrease the influence on the impedance from ambient temperature [20]. A large amount of research has already been reported about the application of MWNTs/polymer composite in developing strain and force sensors in which the sensitive elements are fabricated in bulk size [6,21,22]; however, in this paper, a novel tactile sensor with the sensitive element in thin aligned fibers form is described. Thin fiber in sub-micrometer diameter not only constrains the dispersion of MWNTs and increases their alignment as well as electrospinning progress does [23], but also the drum collector arranges the polymer fibers parallel so that MWNTs in different fibers remain in the same orientation which makes SEs show anisotropy in electricity (see Section 4). The electrical conductivity of sensitive fiber is mainly decided by the distance between the conduct fillers (the conductivity of MWNTs is 104 S/m, whereas that of PU is usually less than 10−12 S/m). However, different from the stretching situation, SMFs in the proposed sensor are compressed in the radial direction as shown in Figure 1a. This drawing also demonstrates a typical dispersion of MWNTs (simplified as long rods) in a PU fiber with an electric passage lined out. The blue lines are the passages along the surface of MWNTs, and red ones are the passages in the polymer that dominate the resistance of the composite fiber. The later passages could be divided into two kinds: the one connects to the outside of the fiber (as 1 and 5 in Figure 1a) and the other connects to the inside of the fiber (as 2, 3 and 4 in Figure 1a). Situation 2 shows that two MWNTs are contacted or within the tunneling distance, while situation 3 shows that two MWNTs are overlapped and four are the separated in the longitudinal direction. Considering the longitudinal distance between MWNT and electrode or between MWNTs

Micromachines Micromachines 2018, 2018, 9, 9, 201 x

33 of of 13 13

MWNT and electrode or between MWNTs in different fibers, situations 1 and 5 could be classified in different fibers, situations 1 and 5 could be classified into situation 3 or 4. Thus, the resistance of one into situation 3 or 4. Thus, the resistance of one fiber could be described as fiber could be described as j

k

m

n

k R  m R  nR Rfiber =j Rcnti   ci oi si

Rfiber = ∑ i 1Rcnti + ∑ i 1 Rci +i ∑ 1 Roi + i 1∑ Rsi i =1

i =1

i =1

(1) (1)

i =1

where Rcnti is the resistance of one MWNT along its surface, Rci is the resistance in situation 2, Roi is where Rcnti resistance4;of along surface,ofRthese resistancerespectively. in situation Because 2, Roi is ci is the situation 3, is Rsithe is situation j, one k, m,MWNT and n are the its amounts resistances situation 3, R is situation 4; j, k, m, and n are the amounts of these resistances respectively. Because si parts are smaller than the latter two by orders, the whole resistance could be estimated the former two the former two parts are smaller than the latter two by orders, the whole resistance could be estimated by formula by formula mm

nn

i= i 11

=11 ii

RR  RRoioi +  fiber ∑RRsisi fiber≈∑

(2) (2)

As As in in Figure Figure 1b, 1b, once once the the fiber fiber is is affected affected by by aa uniform uniform load, load, the the radius radius in in the the yy direction direction will will compress and that in the x direction will elongate. However, because of the high aspect ratio of SMF, compress and that in the x direction will elongate. However, because of the high aspect ratio of SMF, the the analysis. analysis. the length length in in the the zz direction direction will will barely barely change, change, and and it it can can thus thus be be ignored, ignored, simplifying simplifying the This This deformation deformation causes causes two two opposite opposite effects effects on on the the whole whole resistance resistance of of the the composite composite fiber: fiber: on on one one hand, according to the law of resistance, the decrement of the section’s area (the elongation in hand, according to the law of resistance, the decrement of the section’s area (the elongation in the the xx direction direction is is shorter shorter than than the the compression compression in in the the yy direction direction for for Poisson’s Poisson’s ratio) ratio) such such as as in in situation situation 33 will will increase increase the the resistance; resistance; on on the the other other hand, hand, the the two two parts parts in in situation situation 22 will will decrease decrease because because the the distance between two MWNTs in the xy plane becomes smaller whereas the distance in the direction distance between two MWNTs in the xy plane becomes smaller whereas the distance in the zz direction changes owing toto thethe distance variation in changes little, little, which whichpossibly possiblycauses causesdramatic dramaticlocal localresistance resistancechange change owing distance variation the range of quantum-tunnelling effect in case of oriented arrangement of MWNTs. The predominated in the range of quantum-tunnelling effect in case of oriented arrangement of MWNTs. The function of the function two contradictories determines the overall resistance change. When the ratio ofWhen MWNTs predominated of the two contradictories determines the overall resistance change. the is near percolation, the resistance of composite fiber will drop significantly under load. ratio of MWNTs is near percolation, the resistance of composite fiber will drop significantly under load.

Figure 1. (a) A typical dispersion of multi-walled carbon nanotubes (MWNTs) (simplified as long Figure 1. (a) A typical dispersion of multi-walled carbon nanotubes (MWNTs) (simplified as long rods) in a polyurethane (PU) fiber, where the blue line indicates the electrical path along the surface rods) in a polyurethane (PU) fiber, where the blue line indicates the electrical path along the surface of of MWNT and one between twoMWNTs; MWNTs;(b) (b)The Thechange changeofofelectrical electricalpath pathbetween between two two MWNTs MWNTs MWNT and redred one between two when the the fiber fiber is is deformed deformed by by applied appliedforce. force. when

Micromachines 2018, 9, 201

4 of 13

2.2. Structure Design Micromachines 2018, 9, x

4 of 13

There are two layers of aligned composite SMFs in the proposed flexible tactile sensors. Structure Design These2.2. layers are mutually perpendicular and fabricated on different sides of an FPCB (Figure 2). A serial of parallelized Cu strips on the polyimide (PI) filmproposed serve asflexible electrodes segment There are two layers of aligned composite SMFs in the tactilewhich sensors. These the sensing layer. PI and Cu layers are 20 µm and 15 µm thick respectively. The Cu strip is 0.25 mm layers are mutually perpendicular and fabricated on different sides of an FPCB (Figure 2). A serial ofwidth and the central distance (CDoE) is set at 0.5 mm, 1.0 mm, and 2.0 mm prepare sensors parallelized Cu stripsof onelectrodes the polyimide (PI) film serve as electrodes which segment theto sensing layer. and Cu spatial layers are 20 μm and The 15 μm thickof respectively. The Cu strip 0.25 mm width the 3.1. with PI different resolutions. detail the fiber structure willisbe discussed inand Section central distance of electrodes (CDoE) is set at 0.5 mm,resistor 1.0 mm,nets andare 2.0 formed mm to prepare sensors with It is very important that, between Cu strips, parallel to lower the resistance different the spatial resolutions. The detail of thefiber. fiber structure will be in Section 3.1. Iton is the verySMFs and increase measurability of composite As in Figure 2, discussed several cuts are made important that, between Cu strips, parallel resistor nets are formed to lower the resistance layer every two electrodes to separate each SE so as to decrease the coupling effect between and adjacent increase the measurability of composite fiber. As in Figure 2, several cuts are made on the SMFs layer ones. Considering an SE affected by a uniform stress, the impedance measured is every two electrodes to separate each SE so as to decrease the coupling effect between adjacent ones. Considering an SE affected by a uniform stress, the a(l − a)impedance R R0 measured is fiber

0 fiber Rm = (3) 0  fiber +RaR (l a− la) R a fiber Rfiber fiber   Rm (3)   l  a  Rfiber  aRfiber 0 where R fiber is the resistance of the fiber in the loading area, Rfiber is the resistance of the fiber out of ′fiber is the resistance of the fiber in the loading area, Rfiber is the resistance of the fiber out of where Rarea, the loading l and a are the lengths of the SE and diameter of loading area respectively which can the loading area, l and a are the lengths of the SE and diameter of loading area respectively which can also indicate the quantity of fibers. Each SE array of one side can detect the force distribution in one also indicate the quantity of fibers. Each SE array of one side can detect the force distribution in one dimension, and the information about force location and amplitude can be derived through combining dimension, and the information about force location and amplitude can be derived through the responses of SEs from both sides. combining the responses of SEs from both sides.

Figure 2. The schematic demonstration of the proposed flexible tactile sensor with two orthogonal

Figure 2. The schematic demonstration the proposed sensor withcircuit two orthogonal unidirectional composite sub-microfiberof(SMFs) layers onflexible both sidetactile of flexible printed board unidirectional composite sub-microfiber (SMFs) layers both side of flexible printed (FPCB). One SE consisting of parallel SMFs between twoon electrodes is indented with a roundcircuit contactboard (FPCB). of parallel SMFs between two electrodes is indented with a round contact areaOne as inSE theconsisting figure. area as in the figure. 2.3. Signal Acquisition

2.3. SignalFigure Acquisition 3 is the diagram of the data acquisition system. However, the normal multiplex circuits cannot obtain the current signal from the SE because of the high resistance. Thus, the resistance

Figure 3 is the diagram of the data acquisition system. However, the normal multiplex circuits signals of piezoresistive arrays are converted to voltage signals by a specially developed 16-channel cannot obtain the current signal from the SE because of the high resistance. Thus, the resistance signals high resistance measuring device. Then, the signals are sent to the acquisition board (NI-6394, of piezoresistive arrays are converted to voltage signals by adata specially developed 16-channel National Instruments Corporation, Austin, TX, USA) and the are analyzed and displayed on a high resistance measuring device. the signals sent to thethe acquisition board (NI-6394, computer. By changing the Then, mass ratio of fillersare in composite, average impedance between National two Instruments Corporation, Austin, TX, USA) and the data are analyzed and displayed on a computer.

Micromachines 2018, 9, 201

5 of 13

Micromachines 2018, 9, x

5 of 13

By changing the mass ratio of fillers in composite, the average impedance between two electrodes Ω (see could be adjusted the scaletoofthe 109scale Section 3.1). Thus,3.1). the Thus, high resistance measuring device is 9 Ω (see electrodes could betoadjusted of 10 Section the high resistance measuring 8 10 designed to detect resistance variation variation ranging from 1.0 ×from 10 1.0 Ω to× 0.4 device is designed to detect resistance ranging 108 × Ω 10 to 0.4Ω.× 1010 Ω.

Figure 3. Diagram of the data acquisition system for the proposed sensor. Figure 3. Diagram of the data acquisition system for the proposed sensor.

3. Development of Sensor 3. Development of Sensor 3.1. Preparation of Composited Material 3.1. Preparation of Composited Material The MWNTs used in this paper are Dimethylformamide (DMF) based on dispersion with The MWNTs used in this paper are Dimethylformamide (DMF) based on dispersion with 7.8 wt % 7.8 wt % of solid (Timesdisper, TNDDM, Chengdu Organic Chemicals, Chengdu, China), and the PU of solid (Timesdisper, TNDDM, Chengdu Organic Chemicals, Chengdu, China), and the PU is is Elastollan 1185 A from BASF SE (BASF Polyurethanes GmbH, Lemförde, Germany). The densities Elastollan 1185 A from BASF SE (BASF Polyurethanes GmbH, Lemförde, Germany). The densities of of the MWNTs and PU are 1.8 g/cm3 and 1.12 g/cm3, respectively. The outside diameter of the the MWNTs and PU are 1.8 g/cm3 and 1.12 g/cm3 , respectively. The outside diameter of the MWNTs MWNTs is 50 nm and the length is around 10~20 μm. To increase the spinnability of the composite, is 50 nm and the length is around 10~20 µm. To increase the spinnability of the composite, we have we have chosen tetrahydrofuran (THF, Sinopharm Chemical Reagent, Shanghai, China) as the chosen tetrahydrofuran (THF, Sinopharm Chemical Reagent, Shanghai, China) as the solvent so that solvent so that PU is kept at 15 wt % of the solution. To maximize the sensitivity of composite, the PU is kept at 15 wt % of the solution. To maximize the sensitivity of composite, the volume ratio volume ratio of filler and polymer should be close to percolation threshold. This value is converted of filler and polymer should be close to percolation threshold. This value is converted to mass ratio to mass ratio for convenience. for convenience. Differently to other conductive fillers, MWNTs can be treated as long rods with a large aspect Differently to other conductive fillers, MWNTs can be treated as long rods with a large aspect ratio. According to Celzard et al.’s work [24], the excluded volume method has good efficiency in ratio. According to Celzard et al.’s work [24], the excluded volume method has good efficiency in estimating the critical concentration of material in this shape. The critical concentration, which is the estimating the critical concentration of material in this shape. The critical concentration, which is the volume fraction of the filler in composite, can be calculated by volume fraction of the filler in composite, can be calculated by

 Vex Vcyl     1  exp  hVex iVcyl   φc c= 1 − exp  − V  hVee i 

(4) (4)

where V Vcyl cyl is that where isthe thevolume volume of of filler filler and and can can be be derived derived from from the the following following equation equation with with assuming assuming that MWNTs are long rods (l) with hemispherical ends (r): MWNTs are long rods (l) with hemispherical ends (r):

4

Vcyl =  4 πr πr33 +  ππlr lr 22 V cyl 33 represents the average exclude volume of this rod-like filler and it is given by

(5) (5)

Micromachines 2018, 9, 201 9, x Micromachines 2018,

6 of 136 of 13

32 3of this2rod-like represents the average exclude V volume πr  8πr l  πrl 2 filler sin  and it is given by e

3



(6)

32 3 πr + 8πr2 ldegree + πrl 2 hof sinthe γiµfillers and complex to be calculated; (6) hVof ei = μ is an indication the 3orientation however, there are two extreme cases of this volume: = π/4 for randomly aligned fillers and an indication the orientation degree the fillers and complex to as be calculated; µ isparallel = 0 for alignedofones. is the total of excluded volume and same μ: 1.4 for however, there are two extreme cases of this volume: γ = π/4 for randomly aligned fillers and γwhen = 0 the randomly and 2.8 for parallel [25]. Thus, the critical concentration can be as high as 29.53% for parallel aligned ones. is the total excluded volume and same as : 1.4 for randomly ex µ MWNTs are totally paralleled. To search for this volume ratio of the composite SMF, we have and 2.8 for parallel [25]. Thus, the critical concentration can be as high as 29.53% the MWNTs prepared and exploited a series of different solutions for the fabricatingwhen process described in are totally search theiscomposite SMF, we have prepared and Sectionparalleled. 3.2 (CDoETo is 0.5 mm for andthis thevolume width ofratio fiberoffilm 25 mm). The surface resistance is measured exploited series of different solutions for Shanghai the fabricating described in Section 3.2and (CDoE is in by a ahigh resistance meter (ZC-90G, Taiouprocess Electronic, Shanghai, China) shown 0.5 mm and the width of fiber film is 25 mm). The surface resistance is measured by a high resistance Figure 4, and the abscissa is the mass ratio for convenience. It is obvious that the critical concentration meter Shanghai25% Taiou Electronic, Shanghai, China) Figure 4, and Considering the abscissa the is (ZC-90G, located between and 30% mass ratio, i.e., 15.5%and andshown 18.6%involume ratio. is theresistance mass ratio It isforce obvious that thethe critical is locatedisbetween 25% offor SE convenience. decreases when is applied, massconcentration ratio of the composite chosen at 25% in and 30% mass ratio, i.e., 15.5% and 18.6% volume ratio. Considering the resistance of SE decreases order to reach reliable resistance measurement. when force is applied, the mass ratio of the composite is chosen at 25% in order to reach reliable resistance measurement.

Figure 4.Figure The surface of one sensitive element (SE) with(SE) the mass ratiomass varying 15% from to 30%. 4. The resistance surface resistance of one sensitive element with the ratiofrom varying 15%

to 30%.

3.2. Fabrication of Sensitive SMF by Modified Electrospinning 3.2. Fabrication of Sensitive SMF by Modified Electrospinning The electrospinning method is an efficient technique to fabricate fine polymer fibers with a diameter The ranging from severalmethod nanometers micrometers [26]. However, thefine randomness of fibers’ electrospinning is antoefficient technique to fabricate polymer fibers with a arrangement a significant defect ofnanometers this method. overcome this many researchers have diameterisranging from several to To micrometers [26].problem, However, the randomness of fibers’ reported several modified techniques In this paper, a drum this collector [30] many is implemented tohave arrangement is a significant defect[27–30]. of this method. To overcome problem, researchers produce thick several and large fine fiber film so that its resistance could be decreased effectively [31]. reported modified techniques [27–30]. In this paper, a drum collector [30] is implemented to The diagram thelarge fabrication system is that shown in Figure 5. The be pipette with aeffectively 1 mm diameter produce thickof and fine fiber film so its resistance could decreased [31]. outlet is mounted on aofmotion stage which is jogging mm/s 5. with stroke with of 25amm during The diagram the fabrication system is shownatin1 Figure Theapipette 1 mm diameter the fabricating process.on Ana aluminum drum with diameter 75 mm is driven byofa 25 step motor to the outlet is mounted motion stage which is ajogging at 1ofmm/s with a stroke mm during 5000fabricating rpm and theprocess. FPCB isAn attached to its surface withaitsdiameter electrodes axis.by It isa important aluminum drum with of parallel 75 mm to is the driven step motor to to connect all the drum to its increase the effect alignment of the to fibers [29]. It A is syringe 5000 rpm andelectrodes the FPCBtoisthe attached surface with its of electrodes parallel the axis. important pump connected toelectrodes the pipette supplying the composite at 0.5ofmL/h. The[29]. distance to is connect all the toand the drum to increase the effectsolution of alignment the fibers A syringe between theistip and drum 6 cm and 12 kVsupplying voltage is applied. pump connected toisthe pipette and the composite solution at 0.5 mL/h. The distance between the tip and drum is 6 cm and 12 kV voltage is applied.

Micromachines 2018, 9, 201

7 of 13

Micromachines 2018, 9, x

7 of 13

Micromachines 2018, 9, x

7 of 13

Figure 5. 5. The of fabrication fabrication system of aligned aligned SMFs SMFs on on FPCB FPCB based based on on electrospinning electrospinning Figure The diagram diagram of system of device. Number 1 is a syringe pump, 2 is a high voltage generator, 3 is a linear motion stage and 4 is device. Number 1 is a syringe pump, 2 is a high voltage generator, 3 is a linear motion stage and 4 is a a motor-driven aluminum drum with FPCB attached. motor-driven aluminum drum with FPCB attached.

Figure 5. The diagram of fabrication system of aligned SMFs on FPCB based on electrospinning Number 1 is a syringethe pump, 2 is a highof voltage generator, a linear the motion stage and 4 isTherefore, is device. difficult to determine thickness the fiber film3 isduring fabrication. motor-driven aluminum drum FPCB attached. is adifficult to determine thewith thickness of the fiber film during the fabrication. Therefore,

It It electrospinning time is chosen as the factor to control the thickness. After the fabrication on one side electrospinning time is chosen as the factor to control the thickness. After the fabrication on one is finished, FPCB istoturned over andthickness rotated 90° before reattached thefabrication. drum, andTherefore, then the other It the is difficult of the film duringtothe ◦ before side is finished, the FPCBdetermine is turnedthe over and rotated 90fiber reattached to the drum, and then the side electrospinning begins to collect. The photo of a fabricated flexible tactile sensor is shown in time is chosen as the factor to control the thickness. After the fabrication on Figure one side6a. A other side begins to collect. The photo of a fabricated flexible tactile sensor is shown in Figure 6a. scanning electron microscope (SEM; SU8010, Hitachi, Tokyo, Japan) image of SMFs between is finished, the FPCB is turned over and rotated 90° before reattached to the drum, and then the other two A scanning electron microscope (SEM; SU8010, Hitachi, Tokyo, Japan) image of SMFs between two side begins to collect. photo of a fabricated tactile sensor isand shown Figure 6a. A are electrodes in Figure 6b is The clearly shown the detailflexible designed structure, the in collected fibers electrodes in Figure 6b is clearly shown the detail designed structure, and the collected fibers are scanning microscope (SEM; SU8010, Hitachi, Tokyo, Japan) ofregulated SMFs between twomain distributed inelectron an acceptable alignment. However, the structure is notimage totally and the distributed in an acceptable alignment. However, the structure is not totally regulated and the main electrodes in Figure 6b is clearly shown the detailsolution designed structure, and the are reason is the high conductivity of electrospinning which increases thecollected bendingfibers instability of reason is the high conductivity of electrospinning solution which increases the bending instability of distributed in an acceptable alignment. However, the structure is not totally regulated and the main jet [32]. Investigation of the high magnification SEM photo (Figure 6c) revealed to us that the reason is the high of conductivity of electrospinning which increases the bending instability of jet [32]. Investigation the high magnification SEMsolution photo (Figure 6c) revealed to us that the diameters diameters of composite fibers were mainly under 1 μm in the sub-micrometer scale. Also, three jet [32]. Investigation of the high magnification photo (Figure scale. 6c) revealed to us transmission that the of composite fibers were mainly under 1 µm in theSEM sub-micrometer Also, three transmission electron microscope (TEM; CM200, Philips, Amsterdam, The Netherlands) images of a diameters of composite fibers were mainly under 1 μm the sub-micrometer Also, composite three electron microscope (TEM; CM200, Philips, Amsterdam, TheinNetherlands) imagesscale. of a single single composite SMF in Figure 7 shows the typical situation 2, 3 and 4 of MWNTs dispersion electron microscope (TEM; CM200, Philips, Amsterdam, The Netherlands) images of a SMF transmission in Figure 7 shows the typical situation 2, 3 and 4 of MWNTs dispersion described in Section 2.1 described Section 2.1 which dominate the the resistance a force applied.dispersion To protect the singlein composite SMF in Figure 7 shows typical change situationwhile 2, 3 and 4 ofisMWNTs which dominate the resistance change while a force is applied. To protect the sensitive fibers, we have sensitive fibers, we have a polyimide (PI) layer (7413D, Company, described in Section 2.1 applied which dominate the resistance change while a3M force is applied.Maplewood, To protect the MN, applied a polyimide (PI) layer (7413D, 3M Company, Maplewood, USA) Maplewood, with the thickness we have applied a polyimide layer (7413D, 3MMN, Company, MN, of USA)sensitive with thefibers, thickness of 60 μm over the SMF (PI) layer. 60 µm over the SMF layer. USA) with the thickness of 60 μm over the SMF layer.

Figure (a)photo A photo of proposedflexible flexible tactile tactile sensor central distance of electrodes Figure 6. (a) (a)6.A A of proposed proposed sensorwith with0.5 0.5mm mm central distance of electrodes electrodes Figure 6. photo of flexible tactile sensor with 0.5 mm central distance of (CDoE); (b) An SEM image of SMFs between two Cu electrodes; (c) A higher amplification SEM image (CDoE); (b) An SEM image of SMFs between two Cu electrodes; (c) A higher amplification SEM image (CDoE); (b) An SEM image of SMFs between two Cu electrodes; (c) A higher amplification SEM image of SMFs to show the alignment. of SMFs SMFs to to show show the the alignment. alignment. of

Micromachines 2018, 9, 201

8 of 13

Micromachines 2018, 9, x

8 of 13

Figure 7. Transmission electron microscope (TEM) images to show positional relations between

Figure 7. Transmission electron microscope (TEM) images to show positional relations between MWNTs in PU SMF: (a) Contacted MWNTs; (b) Separated MWNTs; (c) Overlapping MWNTs. MWNTs in PU SMF: (a) Contacted MWNTs; (b) Separated MWNTs; (c) Overlapping MWNTs.

4. Test and Discussion of Results

4. Test and Discussion of Results 4.1. Piezoresistivity of Single SE

4.1. Piezoresistivity of Single SE To measure the piezoresistivity of the parallel SMF structure between two electrodes, a polymethylmethacrylate (PMMA) disk with diameter of 10SMF mm isstructure placed on the composite material To measure the piezoresistivity of athe parallel between two electrodes, of the proposed sensor and the SE under the center of the disk is observed. The resistance changes a polymethylmethacrylate (PMMA) disk with a diameter of 10 mm is placed on the composite of material SEs with different fabrication time are measured by the high resistance meter when the readout is of the proposed sensor and the SE under the center of the disk is observed. The resistance changes stable after different weights are placed on this square. The test results drawn in Figure 8a show that of SEs with different fabrication time are measured by the high resistance meter when the readout is the impedance of SMF decreases dramatically when the weight starts to increase (the initial weight stable after different weights are of placed on this square. The test results drawn in adding Figure weight 8a show that is 20 grams) and, after a period fast dropping, it slowly reaches a flat area where the (increased impedance of SMF dramatically when the weight modulus starts to of increase (the initial to over 1000decreases gram) influences it little because the Young’s PU increases greatlyweight is 20along grams) a period of fast dropping, it slowly a flatfabricated area where adding withand, the after increasing of deformation. To fit the resultsreaches from SMFs in 10 min, aweight logarithm is used forinfluences the resistance change: (increased to function over 1000 gram) it little because the Young’s modulus of PU increases greatly along with the increasing of deformation. To fit the results from SMFs fabricated in 10 min, a logarithm y  1.087  0.142 ln x  4.81  10 4 (7) function is used for the resistance change:





According to Fechner’s law [33], for human perception, the relation   between stimulus and y= −1.087 0.142 lncurve x + of 4.81 × 10−4 tactile sensor. This property (7) perception is a logarithm which just fits the−sensitive proposed grants the sensor high sensitivity to distinguish the small differences in low force area and a large range of measurement. According to Fechner’s law [33], for human perception, the relation between stimulus and By changing the electrospinning time, a series of SMF layers different tactile thicknesses is produced perception is a logarithm which just fits the sensitive curve of of proposed sensor. This property and the increase of the thickness (represented by electrospinning time) reduces the change grants the sensor high sensitivity to distinguish the small differences in low force arearate andofa large resistance obviously, as in Figure 8a. The main reason for this reduction could be that the thicker the range of measurement. SMF layer is, the less the decrement ratio of thickness under the same weight would become because By changing the electrospinning time, a series of SMF layers of different thicknesses is produced the strain of the elastic layer that counterbalances the pressure stays the same. and theTo increase of effect the thickness by electrospinning time) reduces the changetorate of study the of parallel(represented resistors networks formed by unidirectional SMFs orthogonal resistance obviously, as in Figure 8a. The main reason for this reduction could be that the thicker the the parallel electrodes, tactile sensors with the direction of SMFs along the parallel electrodes are SMFprepared layer is,and theserved less theasdecrement ratio of comparisons thickness under the same weight would because comparatives. The between resistances of SEs frombecome different the sensors strain ofare the elastic counterbalances the pressure same. arrangements are drawn in layer Figurethat 8b while the SEM images of SMFs stays in twothe different shown in Figure 8c,d. The impedances from counterparts are higher than from the proposed sensor To study the effect of parallel resistors networks formed by unidirectional SMFs orthogonal to and this disparity is maintained at around 150% as the CDoE enlarges, which means the proposed the parallel electrodes, tactile sensors with the direction of SMFs along the parallel electrodes are structure has a persistent effect on decreasing the resistance of SE. resistances This phenomenon be prepared and served as comparatives. The comparisons between of SEs could from different explained as that the MWNTs in the SMFs orthogonal to the electrodes could form the electrical paths sensors are drawn in Figure 8b while the SEM images of SMFs in two different arrangements are shown as described in Section 2.1 whereas electrical paths in the SMFs parallel to the electrodes would in Figure 8c,d. The impedances from counterparts are higher than from the proposed sensor and this mainly depend on situations 1 and 5 which have higher impedances than the rest.

disparity is maintained at around 150% as the CDoE enlarges, which means the proposed structure has a persistent effect on decreasing the resistance of SE. This phenomenon could be explained as that the MWNTs in the SMFs orthogonal to the electrodes could form the electrical paths as described in Section 2.1 whereas electrical paths in the SMFs parallel to the electrodes would mainly depend on situations 1 and 5 which have higher impedances than the rest.

Micromachines 2018, 9, 201 Micromachines 2018, 9, x Micromachines 2018, 9, x

9 of 13 9 of 13 9 of 13

Figure 8. (a) The resistance change of SE with different thickness of SMFs layer under applied weight Figure 8. (a) The The resistance resistance change change of SE SE with with different different thickness of layer under applied weight Figure 8. (a) thickness of SMFs SMFs under applied weight (measured in 10~15 s after weight of placed); (b) The resistance change ratiolayer comparison between SMFs (measured in 10~15 s after weight placed); (b) The resistance change ratio comparison between SMFs (measured in 10~15 s after weight placed); (b) The resistance change ratio comparison between SMFs perpendicular and parallel to electrodes with different CDoE; (c) SEM image of SMFs perpendicular perpendicular and parallel to electrodes with different CDoE; (c) SEM image of SMFs perpendicular to perpendicular andSEM parallel to electrodes with different CDoE; (c) SEM image of SMFs perpendicular to electrodes; (d) image of SMFs parallel to electrodes. electrodes; (d) SEM image of SMFs parallel to electrodes. to electrodes; (d) SEM image of SMFs parallel to electrodes.

4.2. Static and Dynamic Validation Test 4.2. Static and and Dynamic Dynamic Validation 4.2. Static Validation Test Test Experiment tests to validate the prototype flexible tactile sensor can be divided into two kinds Experiment tests to validate the flexible tactile sensor be into two Experiment tests to validate the prototype prototype flexible tactilevalidation sensor can cantest. be divided divided intoimplements two kinds kinds by the equipment setups: static validation test and dynamic The former by the equipment setups: static validation test and dynamic validation test. The former implements by the equipment setups: static validation test and dynamic validation test. The former implements a static indentation test and sliding test with the combination of an auto positioner (VP-25XYZL, aa static test and sliding test with combination of an positioner (VP-25XYZL, static indentation indentation test andand sliding test motion with the the combination of an auto autowhile positioner (VP-25XYZL, Newport, Irvine, CA, USA) a linear stage (LM-600, Newport) the latter performs a Newport, Irvine, CA, USA) and a linear motion stage (LM-600, Newport) while the latter aa Newport, Irvine, CA, USA) andaavibration linear motion stage (LM-600, Newport) while the latter performs performs dynamic indentation test with exciter (V406, Brüel & Kjær, Nærum, Denmark, controlled dynamic indentation test a vibration vibration exciter (V406, Brüel Brüel & & Kjær, Kjær, Nærum, Denmark, controlled controlled dynamic indentation test with with exciter (V406, by Spider-81B controller froma Crystal Instruments, Santa Clara, CA,Nærum, USA) asDenmark, the photos shown in by Spider-81B controller from Crystal Instruments, Santa Clara, CA, USA) as the photos by Spider-81B controller from Crystal Instruments, Santa Clara, CA, USA) as the photos shown shown in in Figure 9. Figure 9. Figure 9.

Figure 9. (a) A photo of static test equipment, where 1 is an auto positioner, 2 is a linear motion stage, Figure 9. (a) A photo of static equipment, where(b) 1 isAan autoof positioner, 2 is aequipment, linear motion stage, 3 is proposed and 4 istest a static force sensor; photo dynamic test where 1 is Figure 9. (a) A sensor photo of static test equipment, where 1 is an auto positioner, 2 is a linear motion stage, 3aisvibration proposedexciter, sensor2and 4 is a static force sensor; (b) A photo of dynamic test equipment, where 1 is is a4dynamic is proposed andtest 4 is equipment, manual positioner. 3 is proposed sensor and is a staticforce forcesensor, sensor;3(b) A photo ofsensor dynamic where 1 is a a vibration exciter, 2 is a dynamic force sensor, 3 is proposed sensor and 4 is manual positioner. vibration exciter, 2 is a dynamic force sensor, 3 is proposed sensor and 4 is manual positioner.

In the static test, parallel SEs on one side of the FPCB are considered. During the static In the static SEsround on one side of different the FPCBdiameters are considered. During the static indentation test, atest, seriesparallel of PMMA disks with is set between indenter and indentation test, a series of PMMA round disks with different diameters is set between indenter and sensor. The resistances of the SMFs under the disk will decrease at the present of indentation force sensor. resistances of the SMFs under disk willshould decrease at the present of indentation forcein and theThe change of impedance between twothe electrodes follow with Equation (3). The results and the change of impedance between two electrodes should follow with Equation (3). The results in

Micromachines 2018, 9, x Micromachines 2018, 9, x

10 of 13 10 of 13

Figure demonstrate that structure parallel a good spatial discrimination Figure 10 10 demonstrate that thethe structure of of parallel SEsSEs hashas a good spatial discrimination in in oneone Micromachines 2018, 9, 201sliding test, a polydimethylsiloxane (PDMS) cylinder is attached to the end 10 of direction. For the of13the direction. For the sliding test, a polydimethylsiloxane (PDMS) cylinder is attached to the end of the indenter to serve as an elastic buffer driven linear motion stage with different velocities. indenter to serve as an elastic buffer andand driven by by thethe linear motion stage with different velocities. A typical output of the parallel SEs under a sliding of 10 mm/s is drawn in Figure 11a, A typical of the parallel SEsone under of are 10 considered. mm/s is drawn in the Figure andand by by In theoutput static test, parallel SEs on side aofsliding the FPCB During static11a, indentation measuring the time differences between the adjacent SEs and considering the distance between them, measuring theof time differences the adjacent and considering the distance between them, test, a series PMMA roundbetween disks with differentSEs diameters is set between indenter and sensor. the sliding speed could be calculated from the signals. The calculation results of varying speed from the sliding speed could be calculated from the signals. The calculation results of varying speed from The resistances of the SMFs under the disk will decrease at the present of indentation force and the 10 mm/s to 50 mm/s are drawn in Figure 11b and compared with the real speed to show the accuracy 10 mm/softoimpedance 50 mm/s are drawntwo in Figure 11b and compared with Equation the real speed to show change between electrodes should follow with (3). The resultsthe in accuracy Figure 10 of the estimation by proposed sensor. of the estimation proposed sensor. demonstrate thatbythe structure of parallel SEs has a good spatial discrimination in one direction. the dynamic test, indentation force is set a sine wave with varying frequencies from In In the dynamic thethe indentation force is set as as a sine with from For the sliding test, atest, polydimethylsiloxane (PDMS) cylinder iswave attached tovarying the end frequencies of the indenter to 5 Hz to 25 Hz while the amplitude is invariant to validate the frequency-domain performance of 5serve Hz toas25an Hz whilebuffer the amplitude is invariant to validate frequency-domain performance of thethe elastic and driven by the linear motionthe stage with different velocities. A typical tactile sensor. Figure 11c draws the filtered signals of one a 0.2 s period under the indentation tactile filtered of one SE SE in ain s period indentation outputsensor. of the Figure parallel11c SEsdraws underthe a sliding ofsignals 10 mm/s is drawn in0.2 Figure 11a,under and bythe measuring the force of 5 different frequencies, and for easy to compare, all the amplitudes are normalized. force of 5 different frequencies, and SEs for easy to compare,the alldistance the amplitudes are normalized. time differences between the adjacent and considering between them, the sliding Figure shows responses of the to the indentation force of different frequencies. From Figure 11d11d shows thethe responses SE SE to the ofof different thethe speed could be calculated from of thethe signals. Theindentation calculation force results varyingfrequencies. speed from From 10 mm/s curves, could find that proposed sensor a good response to the dynamic force with the curves, wewe could find that thethe proposed hashas awith good response to the dynamic force with to 50 mm/s are drawn in Figure 11b andsensor compared the real speed to show the accuracy ofthe the frequency under 25 Hz. frequency 25 Hz. sensor. estimationunder by proposed

Figure 10. The results (filtered by Hz 20 Hz low-pass) of indentation test with different contact disks on Figure 10. by 20 low-pass) of test Figure 10. The Theresults results(filtered (filtered by(a) 20 5Hz low-pass) ofindentation indentation testwith withdifferent differentcontact contactdisks diskson on parallel SEs (CDoE is 1.0 mm): mm diameter; (b) 10 mm diameter. parallel parallel SEs SEs(CDoE (CDoE isis 1.0 1.0 mm): mm): (a) (a)55mm mmdiameter; diameter;(b) (b) 10 10 mm mmdiameter. diameter.

Figure 11. (a) Output of parallel SEs (CDoE is 1 mm) under a sliding of 10 mm/s filtered by 30 Hz Figure Figure 11. 11. (a) (a)Output Outputof ofparallel parallelSEs SEs(CDoE (CDoEisis 11 mm) mm) under under aa sliding sliding of of 10 10 mm/s mm/sfiltered filteredby by30 30Hz Hz low-pass; Calculated speed from time difference of output’s peak between adjacent SEs; low-pass; (b)(b) Calculated speed from thethe time difference of output’s output’s peak between adjacent SEs; low-pass; (b) Calculated speed from the time difference of peak between adjacent SEs; (c) Normalized output of a SE under sine indentations with different frequencies filtered by 30 (c) Normalized Normalized output output of of aa SE SE under under sine sine indentations indentations with with different different frequencies frequencies filtered filtered by by 30 30Hz HzHz (c) low-pass; (d) The responses of the SE to sine indentations with different frequencies. low-pass;(d) (d)The Theresponses responses of ofthe the SE SE to to sine sine indentations indentations with with different different frequencies. low-pass;

Micromachines 2018, 9, 201

11 of 13

In the dynamic test, the indentation force is set as a sine wave with varying frequencies from 5 Hz to 25 Hz while the amplitude is invariant to validate the frequency-domain performance of the tactile sensor. Figure 11c draws the filtered signals of one SE in a 0.2 s period under the indentation force of 5 different frequencies, and for easy to compare, all the amplitudes are normalized. Figure 11d shows the responses of the SE to the indentation force of different frequencies. From the curves, we could find that the proposed sensor has a good response to the dynamic force with the frequency under 25 Hz. 5. Conclusions Composite SMF has both the piezoresistive property of conductive composite and good mechanical characteristics from its sub-micro scale. The proposed structure of unidirectional SMFs orthogonal to the parallel electrodes could decrease the impedances of SEs while increasing the piezoresistive property of them. The fabrication of the tactile sensor proposed in this paper is simple to be realized, and the flexibility of this tactile sensor is obvious, as shown in the photo. The prototype sensors are fabricated and their static and dynamic performances are also tested. The proposed sensor could detect applied static force changing within 10 N and has a minimum sensitive to 0.2 N; it also could measure the speed of sliding in the direction perpendicular to the electrodes. For dynamic testing, this sensor could follow the sine indentation with frequencies from 5 Hz to 25 Hz. However, the surface resistance of SE is still too large (~109 Ω) which produces very small signals that hinder the development of signal acquisition circuits and weaken the capacity of resisting interference. Further research on this flexible tactile sensor should focus on increasing the conductivity of SMF, and also another future work would be to mount the sensor onto a prosthetic and grant closed loop control on the task of grasping subjects. Author Contributions: Weiting Liu conceived and designed the experiments; Xiaoying Cheng performed the experiments and wrote the paper; Xiaodong Ruan and Xin Fu edited the manuscript. Funding: This research was funded by Basic Public Welfare Project of Zhejiang Province, grant number LGJ18E050001; Science Fund for Creative Research Groups of National Natural Science Foundation of China, grant number 51521064; National Natural Science Foundation of China, grant number 51705466. Acknowledgments: The research in this present paper was supported by the Basic Public Welfare Project of Zhejiang Province (LGJ18E050001), Science Fund for Creative Research Groups of National Natural Science Foundation of China (No. 51521064) and the National Natural Science Foundation of China (No. 51705466). Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5. 6.

7. 8.

Kappassov, Z.; Corrales, J.-A.; Perdereau, V. Tactile sensing in dexterous robot hands—Review. Robot. Auton. Syst. 2015, 74, 195–220. [CrossRef] Wang, X.; Dong, L.; Zhang, H.; Yu, R.; Pan, C.; Wang, Z.L. Recent progress in electronic skin. Adv. Sci. 2015, 2. [CrossRef] [PubMed] Chortos, A.; Liu, J.; Bao, Z. Pursuing prosthetic electronic skin. Nat. Mater. 2016, 1–14. [CrossRef] [PubMed] Zou, L.; Ge, C.; Wang, Z.; Cretu, E.; Li, X. Novel tactile sensor technology and smart tactile sensing systems: A review. Sensors 2017, 17, 2653. [CrossRef] [PubMed] Stassi, S.; Cauda, V.; Canavese, G.; Pirri, C.F. Flexible tactile sensing based on piezoresistive composites: A review. Sensors 2014, 14, 5296–5332. [CrossRef] [PubMed] Pyo, S.; Lee, J.-I.; Kim, M.-O.; Chung, T.; Oh, Y.; Lim, S.-C.; Park, J.; Kim, J. Batch fabricated flexible tactile sensor based on carbon nanotube-polymer composites. In Proceedings of the IEEE 2013 Transducers & Eurosensors XXVII: The 17th International Conference on Solid-State Sensors, Actuators and Microsystems (TRANSDUCERS & EUROSENSORS XXVII), Barcelona, Spain, 16–20 June 2013; pp. 1004–1007. Drimus, A.; Kootstra, G.; Bilberg, A.; Kragic, D. Design of a flexible tactile sensor for classification of rigid and deformable objects. Robot. Auton. Syst. 2014, 62, 3–15. [CrossRef] Canavese, G.; Stassi, S.; Fallauto, C.; Corbellini, S.; Cauda, V.; Camarchia, V.; Pirola, M.; Pirri, C.F. Piezoresistive flexible composite for robotic tactile applications. Sens. Actuators A Phys. 2014, 208, 1–9. [CrossRef]

Micromachines 2018, 9, 201

9.

10. 11. 12. 13.

14. 15. 16. 17. 18. 19. 20.

21. 22.

23. 24. 25. 26. 27. 28. 29. 30.

31.

12 of 13

Stassi, S.; Canavese, G.; Cosiansi, F.; Gazia, R.; Fallauto, C.; Corbellini, S.; Pirola, M.; Cocuzza, M. Smart piezoresistive tunnelling composite for flexible robotic sensing skin. Smart Mater. Struct. 2013, 22, 125039. [CrossRef] Kuchibhatla, S.V.N.T.; Karakoti, A.S.; Bera, D.; Seal, S. One dimensional nanostructured materials. Prog. Mater. Sci. 2007, 52, 699–913. [CrossRef] Lu, X.; Zhang, W.; Wang, C.; Wen, T.C.; Wei, Y. One-dimensional conducting polymer nanocomposites: Synthesis, properties and applications. Prog. Polym. Sci. 2011, 36, 671–712. [CrossRef] Ding, B.; Wang, M.; Wang, X.; Yu, J.; Sun, G. Electrospun nanomaterials for ultrasensitive sensors. Mater. Today 2010, 13, 16–27. [CrossRef] Mokhtari, F.; Salehi, M.; Zamani, F.; Hajiani, F.; Zeighami, F.; Latifi, M. Advances in electrospinning: The production and application of nanofibres and nanofibrous structures. Text. Prog. 2016, 48, 119–219. [CrossRef] Wang, Y.R.; Zheng, J.M.; Ren, G.Y.; Zhang, P.H.; Xu, C. A flexible piezoelectric force sensor based on PVDF fabrics. Smart Mater. Struct. 2011, 20, 45009. [CrossRef] Shin, M.; Song, J.H.; Lim, G.H.; Lim, B.; Park, J.J.; Jeong, U. Highly stretchable polymer transistors consisting entirely of stretchable device components. Adv. Mater. 2014, 26, 3706–3711. [CrossRef] [PubMed] Wang, L.; Ding, T.; Wang, P. Thin flexible pressure sensor array based on carbon black/silicone rubber nanocomposite. IEEE Sens. J. 2009, 9, 1130–1135. [CrossRef] Sherman, R.D.; Middleman, L.M.; Jacobs, S.M. Electron transport processes in conductor-filled polymers. Polym. Eng. Sci. 1983, 23, 36–46. [CrossRef] McLachlan, D.S.; Blaszkiewicz, M.; Newnham, R.E. Electrical resistivity of composites. J. Am. Ceram. Soc. 1990, 73, 2187–2203. [CrossRef] Hu, N.; Karube, Y.; Yan, C.; Masuda, Z.; Fukunaga, H. Tunneling effect in a polymer/carbon nanotube nanocomposite strain sensor. Acta Mater. 2008, 56, 2929–2936. [CrossRef] Shih, W.-P.; Tsao, L.-C.; Lee, C.-W.; Cheng, M.-Y.; Chang, C.; Yang, Y.-J.; Fan, K.-C. Flexible temperature sensor array based on a graphite-polydimethylsiloxane composite. Sensors 2010, 10, 3597–3610. [CrossRef] [PubMed] Khan, S.; Tinku, S.; Lorenzelli, L.; Dahiya, R.S. Flexible tactile sensors using screen-printed P(VDF-TrFE) and MWCNT/PDMS composites. IEEE Sens. J. 2015, 15, 3146–3155. [CrossRef] Vatani, M.; Engeberg, E.D.; Choi, J.-W. Detection of the position, direction and speed of sliding contact with a multi-layer compliant tactile sensor fabricated using direct-print technology. Smart Mater. Struct. 2014, 23, 95008. [CrossRef] Dror, Y.; Salalha, W.; Khalfin, R.L.; Cohen, Y.; Yarin, A.L.; Zussman, E. Carbon nanotubes embedded in oriented polymer nanofibers by electrospinning. Langmuir 2003, 19, 7012–7020. [CrossRef] Celzard, A.; McRae, E.; Deleuze, C.; Dufort, M.; Furdin, G.; Marêché, J. Critical concentration in percolating systems containing a high-aspect-ratio filler. Phys. Rev. B 1996, 53, 6209–6214. [CrossRef] Balberg, I. Universal percolation-threshold limits in the continuum. Phys. Rev. B 1985, 31, 4053–4055. [CrossRef] Huang, Z.-M.; Zhang, Y.-Z.; Kotaki, M.; Ramakrishna, S. A review on polymer nanofibers by electrospinning and their applications in nanocomposites. Compos. Sci. Technol. 2003, 63, 2223–2253. [CrossRef] Theron, A.; Zussman, E.; Yarin, A.L. Electrostatic field-assisted alignment of electrospun nanofibres. Nanotechnology 2001, 12, 384–390. [CrossRef] Li, D.; Wang, Y.; Xia, Y. Electrospinning of polymeric and ceramic nanofibers as uniaxially aligned arrays. Nano Lett. 2003, 3, 1167–1171. [CrossRef] Katta, P.; Alessandro, M.; Ramsier, R.D.; Chase, G.G. Continuous electrospinning of aligned polymer nanofibers onto a wire drum collector. Nano Lett. 2004, 4, 2215–2218. [CrossRef] Kim, K.W.; Lee, K.H.; Khil, M.S.; Ho, Y.S.; Kim, H.Y. The effect of molecular weight and the linear velocity of drum surface on the properties of electrospun poly(ethylene terephthalate) nonwovens. Fibers Polym. 2004, 5, 122–127. [CrossRef] Teo, W.E.; Ramakrishna, S. A review on electrospinning design and nanofibre assemblies. Nanotechnology 2006, 17, R89–R106. [CrossRef] [PubMed]

Micromachines 2018, 9, 201

32. 33.

13 of 13

Sun, Z.; Deitzel, J.M.; Knopf, J.; Chen, X.; Gillespie, J.W. The effect of solvent dielectric properties on the collection of oriented electrospun fibers. J. Appl. Polym. Sci. 2012, 125, 2585–2594. [CrossRef] Masin, S.C.; Zudini, V.; Antonelli, M. Early alternative derivations of Fechner’s law. J. Hist. Behav. Sci. 2009, 45, 56–65. [CrossRef] [PubMed] © 2018 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/).