T

JSa

b

c

a

ARRAA

KAFOPSTTU

1

apompp

mrpobpn

o

mT

0d

Sensors and Actuators A 186 (2012) 63– 68

Contents lists available at SciVerse ScienceDirect

Sensors and Actuators A: Physical

jo u rn al hom epage: www.elsev ier .com/ locate /sna

wo axis optoelectronic tactile shear stress sensor

eroen Missinnea,c,∗ , Erwin Bosmanb,c , Bram Van Hoea,c , Rik Verplanckea,c , Geert Van Steenbergea,c ,andeep Kalathimekkada,c, Peter Van Daeleb,c, Jan Vanfleterena,c

Centre for Microsystems Technology, ELIS Department, Ghent University, Technologiepark 914A, B-9052 Gent, BelgiumCentre for Microsystems Technology, INTEC Department, Ghent University, Technologiepark 914A, B-9052 Gent, BelgiumCentre for Microsystems Technology, IMEC, Technologiepark 914A, B-9052 Gent, Belgium

r t i c l e i n f o

rticle history:eceived 30 September 2011eceived in revised form 18 January 2012ccepted 18 January 2012vailable online 2 February 2012

eywords:rtificial skin

a b s t r a c t

Tactile sensors are often used on irregular or moving surfaces, for example as artificial skin on a robotichand or inside a prosthetic socket. In addition to tactile sensors able to detect pressure, there is also aneed to detect shear stress, associated with slippage or friction. Therefore, this paper demonstrates atechnology for fabricating unobtrusive flexible tactile shear sensors using a thin and mechanically strongpolyimide substrate. The operation of the sensor is based on the changing optical coupling between aVertical-Cavity Surface-Emitting Laser (VCSEL) and photodiode. Since this sensor is based on an opticalprinciple, it is less susceptible to environmental influences and electromagnetic interference (EMI). Fur-

lexibleptoelectronicolyimidehear sensoractile sensorwo axis sensor

thermore, a new design of this sensor was developed and tested to enable the detection of both shearforce magnitude and direction (two axis sensor).

© 2012 Elsevier B.V. All rights reserved.

nobtrusive

. Introduction

Tactile sensors are gaining importance in several fields suchs robotics, sports or the medical sector. The possibility of equip-ing robots or physically disabled people with an artificial sensef touch is probably the most appealing application. Further-ore, tactile sensors can be used to measure stresses inside a

rosthetic socket or sports shoes to monitor and optimize usererformance.

In several of these applications, not only pressure sensing (“nor-al stress”), but additionally shear stress sensing plays a significant

ole since this type of stress is associated with friction or slip-age effects. For example, to allow a robotic hand to lift delicatebjects, the amount of slippage determines the grasping force toe exerted. Similarly, the movement of a residual limb inside arosthetic socket, e.g. during walking, causes frictional stress which

eeds to be limited to prevent serious skin injury.

Obviously, the required tactile sensors need to be adjusted forperating on, or in contact with the human body. Firstly, the sensor

∗ Corresponding author at: Centre for Microsystems Technology, ELIS Depart-ent, Ghent University, Technologiepark 914A, B-9052 Gent, Belgium.

el.: +32 92645351; fax: +32 92645374.E-mail address: Jeroen.Missinne@elis.ugent.be (J. Missinne).

924-4247/$ – see front matter © 2012 Elsevier B.V. All rights reserved.oi:10.1016/j.sna.2012.01.038

should be unobtrusive, i.e. thin, flexible and without protrusionswhich may irritate the skin, for example inside a prosthetic socket.Secondly, the sensor operation should not be influenced by thehuman body.

In literature, several types of shear sensors have been reported.Earlier, the operation of these sensors mostly relied on MEMS(Microelectromechanical Systems), fabricated using silicon micro-machining techniques. A typical sensor configuration applies thepiezoresistive effect in these semiconductors to detect mechani-cal stresses [1–3]. Such sensors can be highly linear and sensitivebut the fabrication process is complicated and the used silicon sub-strate is inherently not flexible. Therefore, the focus is currentlyshifting towards polymers, which can be exploited more easily tofabricate flexible sensors. A large amount of such polymer sensorsare based on the simple principle of changing capacitance [4,5], andare consequently also susceptible to electromagnetic interference(EMI) or stray capacitance. This situation is unfavorable since thesensors need to be used in proximity with the human body whichacts as a stray capacitance, influencing the sensor operation.

Systems using optical sensing principles do not suffer from thesenegative effects and therefore a shear sensor is presented based on

the changing optical coupling between a light source and a detector.Furthermore, the system can be implemented using thin and flexi-ble layers resulting in an unobtrusive sensor which can be appliedon the human skin or other irregular surfaces.

64 J. Missinne et al. / Sensors and Actuators A 186 (2012) 63– 68

photodiod e chip

VCSEL chip

Ultra thi n flexibl e VCSE L package

Ultra thi n flexibl e photodiod e package

PDMS transduce r layer 180µm

2

2

paaaFomb(

2

pvViekar

eblmutcaaValwtmutgpttsa

fp

(1) (2) (3)

Fig. 2. Various shear sensor configurations using different photodiode shapes anddifferent initial lateral VCSEL-to-photodiode alignment (as seen from the top). The

either the VCSEL beam profile ˚ (x, y), or the photodiode function

Fig. 1. Principle of the shear sensor (side view).

. Sensor design

.1. Principle

The principle of the sensor is based on the changing optical cou-ling between a Vertical-Cavity Surface-Emitting Laser (VCSEL) and

photodiode depending on their relative lateral displacement. Such sensor is able to record displacements and can be transformed into shear stress sensor using a transducer material, as illustrated inig. 1. In order to make the sensor thin and flexible, the requiredptoelectronic components were thinned and embedded in poly-er foils of about 40 �m thickness. These foils were separated

y a 180 �m thick Polydimethylsiloxane (PDMS) transducer layerSylgard® 184, Dow Corning) resulting in the final sensor topology.

.2. Optical sensor architecture

The modeling of the sensor can be subdivided in two inde-endent studies. The first study involves the photodiode currentariation as a function of the relative lateral displacement of theCSEL and photodiode (“optical response”), while the second study

nvolves the relation between the applied shear stress and lat-ral displacement of the top sensor part when the bottom part isept fixed. Using this mechanical relation, the photodiode currents a function of the applied shear force is obtained (“mechanicalesponse”).

The mechanical response depends on the dimensions and prop-rties of the transducer layer. It was found that the mechanicalehavior of this material, the relation between shear stress and

ateral displacement, was linear for small displacements (experi-entally verified up to 100 �m). The optical response was studied

sing an analytical sensor model based on simplified characteris-ics of the optoelectronics. Therefore, the photodiode sensitivityurve was considered to be linear, uniform and independent of thengle of incidence, which is valid since the VCSEL beam exhibits

small divergence angle (typically 3–8◦, single sided). The usedCSELs are multi-transverse mode emitting (250 �m pitch 1 × 4rray, ULM Photonics [6]) but their beam profile was neverthe-ess considered to be Gaussian. From the experimental results it

as shown that this simplification is valid when the vertical dis-ance between VCSEL and photodiode is limited to a few hundred

icrometer, since the photodiode is then averaging out any irreg-larities in the beam profile. The Gaussian beam is specified by theotal emitted optical power and opening angle, defined as the sin-le sided angle wherein 50% of the optical power is confined. Thesearameters depend on the driving current and were measured [7]o serve as an input for the analytical model. In the remainder ofhis paper, a VCSEL driving current of 5 mA is considered, corre-ponding with an opening angle of 6◦ and a total emitted power ofbout 1.7 mW.

The optical response � (x, y), i.e. the photodiode current as aunction of the lateral (x, y) position of the VCSEL relative to thehotodiode can be calculated as a convolution of a two-dimensional

small red circles represent the VCSELs and the blue structures represent the photo-diodes. (For interpretation of the references to color in this figure legend, the readeris referred to the web version of this article.)

function representing the photodiode, PD(x, y), and another func-tion representing the VCSEL beam, ˚d(x, y):

� (x, y) = (˚d ∗ PD)(x, y) =∫ ∫

˚d(x′, y′)PD(x − x′, y − y′)dx′dy′

The photodiode function PD(x, y) evaluates in either a con-stant value or 0, for every (x, y) respectively inside or outsidethe photosensitive area. The VCSEL beam function ˚d(x, y) is atwo-dimensional Gaussian function, representing the intensity dis-tribution in the beam, considered at a specific distance d from theemitting surface. Since the VCSEL beam is diverging, ˚d(x, y) rep-resents a broader function for larger values of d (comparable to aGaussian distribution with a large variance) and a narrow functionfor small values of d (comparable to a Gaussian distribution with asmall variance).

Evaluating this equation for every possible relative lateralposition (x, y) of the VCSEL yields the general, two-dimensionalresponse � (x, y). However, typically only the response correspond-ing with a displacement in a certain direction is needed, whichcan be obtained by evaluating along a specific line, e.g. the x-axis,resulting in the x-directional response � x(x) = � (x, 0). Furthermore,� (0, 0) or � x(0) represents the sensor response when idle (no shearforce applied) and hence corresponds to the optical power incidentto the photodiode for the initial lateral VCSEL-to-photodiodealignment. If this mutual initial lateral alignment is changed with acertain amount, the sensor response curve is shifted with the sameamount.

The VCSEL and photodiode are also vertically separated by acertain distance d, see Fig. 1. This optimum vertical VCSEL-to-photodiode separation d was previously determined to be 200 �m[8]. Since both the photodiode and VCSEL chip are covered with aprotective polymer layer (see Fig. 1 and Section 3.), this separationdistance corresponds with a transducer layer thickness of 180 �m.Additionally, in [9], it was simulated that normal forces have a verylimited influence on the sensor response to shear force.

In [8], a first generation flexible shear sensor was proposed illus-trating the optoelectronics integration technology. However, thissensor was constructed using a VCSEL initially centered above a100 �m diameter circular photodiode (see Fig. 2, configuration 1)and therefore only the total displacement or shear force magnitudecan be resolved. Due to the symmetric configuration, it is not possi-ble to distinguish the direction in which the VCSEL is displaced. Thiscan also directly be deduced from the calculated two-dimensionalsensor response � (x, y), plotted in Fig. 3: when the VCSEL is movingaway from its initial position (0, 0), the photodiode signal decreases,but the direction of displacement cannot be determined.

To obtain the desired two axis sensor which is able to recordboth magnitude and direction of displacement, the sensor response� (x, y) needs to be modified. This can be achieved by changing

dPD(x, y). The VCSEL beam profile cannot easily be changed unlessa special filter, lenses or a different light source is used. However,the photodiode function can easily be changed by adapting the

J. Missinne et al. / Sensors and Act

Fig. 3. Sensor response � (x, y) corresponding with a 100 �m round photodiode(VCSEL initially aligned above the middle of the photosensitive area). (For interpre-tw

sfac2itvrp

iacitisisetmrput

Ft(r

ation of the references to color in this figure legend, the reader is referred to theeb version of this article.)

hape of the photosensitive area. When selecting square detectorsor example, the sensor can be used to measure both magnitudend direction of displacement. Fig. 2 illustrates 2 possibilities foronstructing such a two axis sensor. The first design (Fig. 2(2)) uses

VCSELs and 2 photodiodes; the leftmost VCSEL–photodiode pairs used for recording variations in the x-direction, independent ofhe rightmost VCSEL–photodiode pair, that is used for recordingariations in the y-direction. However, these 2 photodiodes can beeplaced by a single photodiode (see Fig. 2(3)), provided that thehotodiode and the 2 VCSELs are driven alternately.

Fig. 4 shows the calculated sensor response � (x, y) correspond-ng with such a VCSEL–photodiode (250 �m × 250 �m square) pair,ble to record variations in the y-direction. The dashed contourorresponds with the location of the photodiode and the red dotndicates the initial position of the VCSEL on the southern edge ofhe photodiode, i.e. (0, 0). Within a substantial region around thenitial VCSEL position (i.e. sensor idle state), the value of the sen-or response (see colored scale) only changes due to displacementsn the y-direction and not due to displacements in the x-direction,ince the VCSEL is moving orthogonally or parallel to the detectordge respectively. Similarly, when initially aligning a VCSEL abovehe western edge of the photodiode, only x-directional displace-

ents influence the corresponding sensor signal. By combining theesponse corresponding with these 2 different VCSEL–photodiodeair situations, a two axis sensor is obtained (either using config-

ration (2) or (3) from Fig. 2). Furthermore, it can be noticed thathe size of the photodiodes determines the sensor dynamic range.

ig. 4. Sensor response � (x, y) corresponding with a 250 �m × 250 �m square pho-odiode (VCSEL initially aligned above the southern edge of the photosensitive area).For interpretation of the references to color in this figure legend, the reader iseferred to the web version of this article.)

uators A 186 (2012) 63– 68 65

3. Fabrication

The fabrication of the shear sensor is based on the packaging ofthinned optoelectronic chips in polymer foils and combining thesefoils with a deformable transducer layer, as depicted schematicallyin Fig. 5 and described in this section.

3.1. Packaging of optoelectronics in flexible polymers

An earlier reported process [7,8] consisted of embedding opto-electronic chips in an SU-8 layer, with a thin PI-2525 polyimidelayer on either side for mechanical reinforcement. This results inan easy fabrication process, but due to the different coefficientsof thermal expansion (CTE) of the used materials, internal stressesexist in the final flexible package. The most critical is the regionaround the embedded chip due to the presence of various materialswith different CTE’s: a Gallium Arsenide (GaAs) chip (6 ppm/◦C), acopper heatsink (17 ppm/◦C) and the PI-2525 (40 ppm/◦C) and SU-8(52 ppm/◦C) embedding layers.

This local accumulation of stresses results in mechanical defor-mation, possibly yielding a small enclosed air bubble at the locationof the chip when bonding the 2 PDMS layers during the final fabri-cation step of the sensor (see Section 3.2). Therefore, an alternativeprocess was established to embed optoelectronics using only poly-imide and no SU-8 as embedding layers. Furthermore, PI-2611was selected instead of PI-2525 (both from HD Microsystems [10])because of its low CTE (3 ppm/◦C) which matches the CTE of theGaAs chip better.

The final process flow is illustrated in Fig. 5. To support the flex-ible polyimide layers during processing, a temporary glass carrierwas used. By coating only the edges of this glass substrate with aprimer, the flexible stack can be cut out and released afterwards.

First, a 5 �m polyimide PI-2611 layer was spin-coated followedby sputtering a 1 �m copper seed layer. This thin layer was thenelectro-plated up to a final thickness of 5 �m and a heatsink wassubsequently defined using photolithography. Then, a second 5 �mlayer of PI-2611 was spin-coated and a cavity was laser ablated inthis layer on top of the heatsink (KrF Excimer laser (� = 248 nm)).The thinned optoelectronic chip (thickness 20 �m) [7] was placedin this cavity using a thermally conductive glue (U 8449-9, Nam-ics Corporation). After thermally curing of the glue, the chip wascovered with another PI-2611 layer and vias were ablated to thecontact pads, using the KrF Excimer laser. The vias were metalizedand simultaneously an electrical fan-out was provided by sput-tering a 1 �m copper layer. After lithographically patterning thecopper, a final 5 �m layer of PI-2611 was spin-coated to finish theprocess. At this point, the polyimide foils with embedded optoelec-tronics were not yet released, to facilitate further processing of theshear sensor, as described below.

3.2. Fabrication of the shear sensor

The next step in the fabrication process of the shear sensor con-sisted of applying a transducer layer between 2 polyimide foilswith embedded components, one with a VCSEL and another with aphotodiode chip.

Therefore, a transducer material (Sylgard® 184, Dow Corning)was firstly spin-coated on 2 such polymer foils with embed-ded optoelectronics. To obtain a final layer thickness of 180 �m,the material was spin-coated at 700 rpm during 60 s, yielding a90 �m thick layer on each polymer foil. Furthermore, a primerwas used (1200 OS, Dow Corning) to achieve a strong adhesion

of the Sylgard® 184 on the polymer package. This Sylgard® 184layer was then kept at room temperature on a leveled platform for24 h and subsequently thermally cured on a hotplate for 60 minat 60◦C. Finally, both foils were aligned and bonded using plasma

66 J. Missinne et al. / Sensors and Actuators A 186 (2012) 63– 68

e emb

s2s

5tftt

sms

3

wuauditsht

Fs

Fig. 5. Process flow for fabricating the shear sensor based on th

urface activation of the Sylgard® 184 [11] (Diener Pico, 0.8 mbar,4 s, 190 W 40 kHz generator, gas used: air), yielding the final sen-or.

Since the chip is thicker (20 �m) than the depth of the cavity �m, the polyimide stack is a locally thicker above the chip becausehe polyimide adopts the profile of the underlying layer. However,or the fabrication of the sensor, this is not a problem, because thehicker PDMS layer applied on top of the polyimide stack levelshese non-uniformities.

Fig. 6 shows the results of the fabrication process. As can beeen in Fig. 6(a), the resulting sensor is very flat because of the lowechanical stress level in the polyimide layers. Fig. 6(b) demon-

trates the mechanical flexibility of the sensor.

.3. Fabricated prototypes

Two generations of sensors were evaluated. The first generationas based on using circular and the second generation based onsing square photodiodes, corresponding to the 2 different sensorrchitectures. The sensors of the first generation were fabricatedsing embedded thin optoelectronics in flexible polyimide foils, asescribed above, demonstrating the technology to obtain mechan-

cally flexible sensors. Furthermore, a prototype of a sensor from

he second generation was fabricated to validate the two axis shearensor design. However, the available square photodiode bare diesave one top contact (anode) and a bottom contact (cathode). Whenhinning down the chip, the bottom contact is removed and needs

ig. 6. Fabricated prototypes: (a) flexible shear sensor after fabrication (first generationubstrate).

edding of thin optoelectronic chips in a flexible polyimide foil.

to be reapplied afterwards, which makes the polyimide based fab-rication process presented above more complicated. Nevertheless,work is currently in progress to adapt the required steps so that aflexible version of the square photodiode based sensor can also befabricated. Currently, the square photodiode was integrated on topof an FR-4 (rigid) substrate, using a cavity in thick SU-8 build-uplayers to avoid the need for thinning the chip. The bottom cath-ode of the photodiode was electrically contacted on a layer ofcopper which was provided underneath this SU-8 layer, while thetop anode was contacted using the microvia technology. Finally, aVCSEL chip packaged in a polyimide foil was aligned with the pho-todiode substrate, using the same process as described above toapply a Sylgard® 184 transducer layer between both components.The resulting sensor sample is shown in Fig. 6(c).

4. Experimental results and discussion

4.1. First design based on round photodiode

The first generation sensor was tested using a Dage Series 4000bondtester with BS100 cartridge to apply a known displacementand measure the resulting shear force. A small cylinder was gluedon the sensor top surface and the flexible sensor itself was glued on

a rigid substrate to facilitate application of the lateral displacementduring testing.

Simultaneously, the VCSEL was excited with a 5 mA cur-rent source while the photodiode was connected with a Source

); (b) sensor flexibility; (c) second generation prototype (implemented using rigid

J. Missinne et al. / Sensors and Actuators A 186 (2012) 63– 68 67

0

0.3

0.6

0.9

1.2

1.5Photodiodecurrent[mA]

0 20 40 60 80 100 120Lateral displacement [μm]

simulationmeasurement

Fig. 7. Sensor response corresponding with a 100 �m round photodiode: pho-tr

Ma

frotglFaact

4

tpTnuam(S

Fr

ocurrent as a function of the lateral VCSEL-to-photodiode displacement (“opticalesponse”).

easure Unit (SMU) to inversely polarize the component at −2 Vnd record the photocurrent.

Fig. 7 shows the measured and simulated photocurrent as aunction of lateral displacement and Fig. 8 the measured photocur-ent as a function of the applied shear force. The initial force offsetn the horizontal axis was caused by the threshold needed to movehe cylinder due to non perfect alignment with the shear tool. Bothraphs are clearly similar owing to the linear relationship betweenateral displacement and shear force. The simulated curve shown inig. 7 matches well with the experiments and can be considered as

cross section of the two-dimensional response graph of Fig. 3, in certain direction. To facilitate the comparison with the simulatedurve, only a limited number of measured points are displayed inhis graph.

.2. Two axis design based on square photodiode

The operation principle of the second generation sensor wasested by recording the photocurrent as a function of lateral dis-lacement, applied under different angles �, as defined in Fig. 4.herefore, a slightly different setup was constructed since it wasot possible to accurately control the angle of lateral displacementsing the Dage bondtester. Instead, the sensor was mounted on

precision rotational stage while a pushing tip fixed on anotherotorized precision stage was used to apply lateral displacements

see Fig. 9). Similarly as described above, a current source and aMU were used to interface the VCSEL and photodiode electrically.

0

0.3

0.6

0.9

1.2

1.5

Photodiodecurrent[mA]

2 3 4 5Applied shear force [N]

ig. 8. Sensor response corresponding with a 100 �m round photodiode: photocur-ent as a function of the applied shear force (“mechanical response”).

Fig. 9. Setup for characterizing the two axis sensor design.

Fig. 4 illustrates the configuration that was tested: a sensor wasconstructed using a square photodiode and a VCSEL was initiallyaligned on the southern edge of the detector.

Since the current setup was not equipped with a force sensor,only the sensor response curves as a function of displacement wererecorded. However, characterization of the first generation sensorshowed a linear relation between lateral displacement and shearforce for the Sylgard® 184 transducer material.

The motorized stage with pushing tip was programmed to applya lateral displacement incrementally onto the sensor and the corre-sponding photocurrent was recorded. This procedure was repeatedand each time the sample was rotated 10◦. The resulting curves areplotted in Fig. 10. It can be seen that the sensor signal is nearlyinvariant when � = 0◦ since the VCSEL is moving parallel to theedge of the photosensitive area (parallel to the x-axis). Contrar-ily, when � = 90◦, the change in signal is maximum since only ay-component is present in the total displacement. For other valuesof �, the total displacement vector is composed of an x- and y-component, resulting in curves with smaller slopes. This is exactlythe desired behavior since this configuration should only be sensi-tive to variations in the y-direction. By using another VCSEL alignedon the western edge of the photosensitive surface, a configura-tion only sensitive to variations in the x-direction is obtained anda combination of both yields a two axis sensor design, see Fig. 2(2)or (3). When using configuration (2), the photocurrent from theleftmost photodiode directly yields the signal corresponding tothe x-component, while the rightmost photodiode directly yields

the signal corresponding to the y-component of the displacement.However, if configuration (3) is used, the photodiode and 2 VCSELsneed to be driven in a scanning operation mode. During a first time

0

0.3

0.6

0.9

1.2

Photodiodecurrent[mA]

0 50 100 150Lateral displacemen t [μm]

30º

20º

10º

χ = 0º

Fig. 10. Sensor response as a function of angle of lateral displacement (square pho-todiode).

6 nd Act

sVadtt

icaeotmabs

5

flsicbe(edca

A

pvPM(

R

[

[

technologies, especially in wearable electronics technologies. As a project manager

8 J. Missinne et al. / Sensors a

lot, only VCSEL 1 is turned on, while during a second time slot, onlyCSEL 2 is turned on. In each time slot, the photodiode current ist least recorded once. Depending on the application, the time sloturation can be chosen to set the operation frequency and/or mul-iple measurements per time slot can be performed for averagingo increase sensor accuracy.

It can be seen that Fig. 10 slightly differs from the ideal behav-or as predicted by the simulations. Firstly, the initial photodiodeurrent is lower than expected because the VCSEL was slightly mis-ligned during fabrication and not exactly positioned above thedge of the detector. Additionally, there is an interference-like,scillating signal superimposed on the sensor response curves. Dueo this phenomenon, the sensor accuracy is limited to about 3 �m

icrometer of lateral displacement. In order to improve the sensorccuracy, this unwanted effect is currently under investigation. It iselieved to be caused by reflections of laser light at the photodiodeurface.

. Conclusions

A technology has been demonstrated to fabricate unobtrusiveexible optical tactile shear sensors using a thin and mechanicallytrong polyimide substrate, with possible applications in prosthet-cs. The principle of the sensor is based on the changing opticaloupling between a VCSEL and a photodiode, which are separatedy a deformable transducer layer. In this paper, a new sensor designnabling the detection of both shear force magnitude and directiontwo axis sensor) was developed and experimentally tested. How-ver, work is still ongoing to implement this new two axis sensoresign in the proposed flexible polyimide technology and to furtherharacterize and optimize the sensor behavior in terms of normalnd shear force cross-sensitivity, accuracy and repeatability.

cknowledgments

This work was partially conducted in the framework of therojects FAOS (funded by the Institute for the Promotion of Inno-ation by Science and Technology (IWT), Flanders, Belgium) andhosfos (funded within the EU-FP7 program). The work of J.issinne is supported by the Research Foundation – Flanders

FWO-Vlaanderen) under a Ph.D. fellowship.

eferences

[1] Characterization of a novel hybrid silicon three-axial force sensor, Sens. Actu-ators A: Phys. 123–124 (2005) 249–257.

[2] M.C. Hsieh, Y.-K. Fang, M.S. Ju, J.-J. Ho, S.F. Ting, Development of a newcontact-type piezoresistive micro-shear-stress sensor, in: Proceedings of theSymposium on Design, Test, Integration, and Packaging of MEMS/MOEMS, vol.4755, 2002, pp. 285–295.

[3] T. Mei, W.J. Li, Y. Ge, Y. Chen, L. Ni, M.H. Chan, An integrated MEMS three-dimensional tactile sensor with large force range, Sens. Actuators A: Phys. 80(2000) 155–162.

[4] K. Sundara-Rajan, A. Bestick, G. Rowe, G. Klute, W. Ledoux, A. Mamishev, Capac-itive sensing of interfacial stresses, in: Proc. IEEE Sens., 2010, pp. 2569–2572.

[5] M.-Y. Cheng, C.-L. Lin, Y.-T. Lai, Y.-J. Yang, A polymer-based capacitive sens-ing array for normal and shear force measurement, Sensors 10 (2010)10211–10225.

[6] Ulm-Photonics, Products. http://www.ulm-photonics.com/ (accessed 2011).[7] E. Bosman, J. Missinne, B. Van Hoe, G. Van Steenberge, S. Kalathimekkad, J. Van

Erps, I. Milenkov, K. Panajotov, T. Van Gijseghem, P. Dubruel, H. Thienpont, P.

Van Daele, Ultrathin optoelectronic device packaging in flexible carriers, IEEEJ. Sel. Top. Quant. 17 (2011) 617–628.

[8] J. Missinne, E. Bosman, B. Van Hoe, G. Van Steenberge, S. Kalathimekkad, P. VanDaele, J. Vanfleteren, Flexible shear sensor based on embedded optoelectroniccomponents, IEEE Photonics Technol. Lett. 23 (2011) 771–773.

uators A 186 (2012) 63– 68

[9] J. Missinne, E. Bosman, B. Van Hoe, G. Van Steenberge, P. Van Daele, J. Van-fleteren, Embedded flexible optical shear sensor, Proc. IEEE Sens. (2010)987–990.

10] HD Microsystems, Products. http://hdmicrosystems.com/HDMicroSystems/en US/products/index.html (accessed 2011).

11] S. Bhattacharya, A. Datta, J. Berg, S. Gangopadhyay, Studies on surface wet-tability of poly(dimethyl) siloxane (PDMS) and glass under oxygen-plasmatreatment and correlation with bond strength, J. Microelectromech. Syst. 14(2005) 590–597.

Biographies

Jeroen Missinne received the Master of Science degree in Electrical Engineeringin 2007 and his PhD in Electrical Engineering in 2011, both at Ghent University,Belgium. He is currently employed as post-doc research engineer at the CMST(Center for MicroSystems Technology), an IMEC-affiliated research lab at GhentUniversity. His research involves polymer waveguides, the embedding of opticalcomponents and in particular the development of embedded optical sensors.

Erwin Bosman was born in Asse, Belgium, in 1980. He graduated in 2004 as anelectrical engineer (Master of Science) in telecommunication and received his PhDin electrical engineering in 2010, both at Ghent University, Belgium. He is currentlyemployed as post-doc research engineer at the CMST (Centre for Microsystems Tech-nology), an IMEC-affiliated research lab at Ghent University. His research focuses onoptical interconnections, integrated optical sensors and electronic chip packages. Heis author or co-author of more than 40 publications (reviewed papers and conferenceproceedings) in the field and co-inventor of one patent.

Bram Van Hoe received the MSc degree in Electrical Engineering in 2008, at theFaculty of Engineering, Ghent University, Belgium. Afterwards, he joined the Centrefor Microsystems Technology (CMST), an IMEC associated lab at Ghent Universitywhere he is currently working towards a PhD on the development and integrationof optical sensors and the corresponding optoelectronic driving units in flexible andstretchable foils. Within this framework, he is also involved in the EC-FP7 fundedproject Phosfos, Photonic Skins for Optical Sensing.

Rik Verplancke received the MSc degree in Electrical Engineering in 2007 at theFaculty of Engineering, Ghent University, Belgium. Since then, he is working towardsa PhD at the Centre of Microsystems Technology (CMST), affiliated with IMEC. Thesubject of his research concerns the development of a generic technology platformfor integration of elastic electronics and microfluidics.

Geert Van Steenberge graduated in Electrical Engineering in 2002 and received hisPhD in Engineering Sciences in 2006, both at Ghent University, Belgium. He is cur-rently employed as research engineer at CMST, an IMEC associated lab at GhentUniversity, and is responsible for the activities on optical sensors, interconnects,and laser technologies. He is involved in several EC-FP7 funded projects, includ-ing PHOSFOS, FAST2LIGHT, ACTMOST, IMPROV, and FIREFLY. Geert Van Steenbergeauthored or co-authored 11 SCI-stated journal papers, and more than 40 publica-tions in international conference proceedings.

Sandeep Kalathimekkad was born in Kozhikode, India, in 1984. He received hisMSc degree in Physics from Cochin University of Science and Technology, India, in2007, MSc in Optics and Photonics from Friedrich-Schiller University Jena, Germanyin 2008 and MSc in Applied Photonics from Imperial College London in 2009. He iscurrently working towards the PhD degree in electrical engineering at the Centrefor Microsystems Technology at Ghent University. His research interests includeintegration of waveguides and other optical components on flexible polymer foils,and coupling with sources and detectors.

Peter Van Daele received the PhD degree in electrical engineering from Ghent Uni-versity, Gent, Belgium, in 1988. He became a permanent member of staff of theInteruniversity Microelectronics Center (IMEC) at the Department of InformationTechnology of Ghent University, where he was responsible for research on the pro-cessing of III–V optoelectronic devices. Since 2001, his work has focused more onoptical packaging and optical interconnections, with emphasis on coupling to fiberarrays, integration on printed circuit boards, and the use of laser processing tech-niques. In 1993, he also became Part-Time Professor. He is author or co-authorof approximately 200 publications in the field of optoelectronic components andtechnology.

Jan Vanfleteren obtained his PhD in electronic engineering from Ghent University,Belgium, in 1987. He is currently a senior engineer at the IMEC-CMST group andis involved in the development of novel interconnection, assembly and substrate

for CMST he has a long standing experience in co-ordination and co-operation inEC funded projects. In 2004 he was appointed as part time professor at the GhentUniversity. He is a member of IMAPS and IEEE and (co)-author of over 200 papers ininternational journals and conferences and he holds 14 patents/patent applications.