Pa

Ta

b

a

ARRAA

KPPTLMC

1

bicllttnb[

dfhimd

s

0d

Sensors and Actuators A 177 (2012) 87– 92

Contents lists available at SciVerse ScienceDirect

Sensors and Actuators A: Physical

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

atterning piezoelectric thin film PVDF–TrFE based pressure sensor for catheterpplication

ushar Sharmaa,∗, Sang-Soo Jea, Brijesh Gillb, John X.J. Zhanga

Biomedical Engineering, University of Texas at Austin, Austin, TX, USAUniversity of Texas Health Science Center at Houston, Houston, TX, USA

r t i c l e i n f o

rticle history:eceived 23 May 2011eceived in revised form 13 August 2011ccepted 22 August 2011vailable online 18 September 2011

a b s t r a c t

We report a thin PVDF–TrFE (polyvinyledenedifluoride–trifluoroethylene) copolymer film pressure sen-sor, fabricated using standard lithography process for cost-effective batch process, film uniformity,and high resolution of polymer patterning. PVDF–TrFE copolymer, a semi-crystalline material, wasspin-coated into thin films (1 �m thick) to tap the near �-phase formation. Pressure measurementsdemonstrated that the thin film (1 �m) show better performance compared to thick film (6 �m) with

eywords:iezoelectricityressure sensorhin film PVDF–TrFEithographyicrofabrication

atheter

no electrical poling or mechanical stretching. Thin film devices showed higher � phase formation fromRaman spectroscopy measurements, which translate into high piezoelectricity for sensing. The sensorscan operate over a broad pressure range of 0–300 mmHg, with fast recovery time of 0.17 s, suitable forreal time flow measurements in catheter applications.

Published by Elsevier B.V.

. Introduction

Balloon occlusion of a major blood vessel is commonly used foroth diagnostic and therapeutic purposes. Every year, clinicians

n the U.S. alone place more than 1.5 million pulmonary arteryatheters for the purpose of hemodynamic monitoring [1]. Bal-oon occlusion of the aorta has emerged as a tool for controllingife-threatening hemorrhage from the pelvis and lower extremi-ies [2]. Currently, the pressure of the inflated balloon is left tohe surgeon’s experience or costly contrast based imaging tech-iques and is not dynamically controlled. Over inflation of thealloon results in a rare but lethal rupture of the occluded artery3].

Patients with massive bleeding display rapidly changing hemo-ynamics and blood pressure. Hence the proper inflation pressureor successful occlusion changes with the patient’s physiology. Aighly compact pressure sensor would allow development of an

nterventional catheter that includes multiple points of pressureeasurement, including inflation pressure as well as upstream and

ownstream blood pressure.Over past decades, many types of microfabricated pressure sen-

ors have been developed, and they can be categorized into three

∗ Corresponding author. Tel.: +1 512 475 6872; fax: +1 512 232 4275.E-mail address: tusharsharma@utexas.edu (T. Sharma).

924-4247/$ – see front matter. Published by Elsevier B.V.oi:10.1016/j.sna.2011.08.019

major groups, depending on their operating principles: capacitive,piezoresistive, and piezoelectric sensors [4–6]. Capacitive sensorsare most commonly adopted in the clinical setting due to relativesimplicity and less expensive fabrication, their high scalability forminiaturization, high sensitivity, and low power consumption, butare mostly applied to low pressures and require complex readoutcircuitry. Piezoresistive sensors, successfully landed into industry,offer great mechanical stability but they have some drawbacks,for instance, a significant power requirement, large temperature-dependence offset, nonlinearity, long-term instability in dynamicfield conditions and impart undesirable stiffness to catheters.Piezoelectric sensors are commonly employed for the measure-ment of highly dynamic pressures.

The need for a pressure sensor on a catheter is not new anddates back to the 1970s [5,7]. So far, silicon based piezoresis-tive and capacitive pressure sensors have been employed for thisapplication but with limited success due to above-mentioned rea-sons. Piezoelectric thin film transducer shows great promise as theimplantable and biocompatible pressure sensors, implemented ona catheter for minimally invasive surgery.

In this paper, we design, microfabricate and characterize piezo-electric PVDF–TrFE (polyvinylidene fluoride–trifluoroethylene)

based pressure sensor which can be later integrated with catheterfor intravascular measurements. Recently, PVDF has gainedincreasing attention for the development of ultra-compact pres-sure sensors [8,9]. However, these sensors comprise of thick PVDF

88 T. Sharma et al. / Sensors and Actuators A 177 (2012) 87– 92

Fig. 1. The unit cell of bulk PVDF in � (top) and � (bottom) phases, combiningfluorine (blue sphere), carbon (green sphere), and hydrogen (gray sphere). The �phase is antipolar, since the dipoles in the alternate chains cancel, while in the �pi

fief

2

oitcp

Phcoocewt[

iuecstb

V

wdagt

oFs

2004006008001000120014001600

Raman Shift (cm-1 )

α

1 μm cured

1 μm uncured

Abs

orba

nce

(a.u

.)

PVDFPVDF αβ

PVDF

β6 μm uncured

β

6 μm curedPVDF

hase they are parallel to each other. (For interpretation of the references to colorn this figure legend, the reader is referred to the web version of the article.)

lms with thickness of 10 �m and above. The present study is anxtension of our previous work [10] on pressure sensor fabricatedrom 1 �m thin PVDF–TrFE films.

. Sensor materials: thin film PVDF

Piezoelectric PVDF, a semicrystalline polymer, may exhibit anyf its four phases, �, �, �, and �. Amongst these phases, � phase ismportant since it has largest effective dipole moment [11]. The keyo achieving high piezoelectricity of the PVDF polymer is forming �-rystalline phase, which significantly improves sensor and actuatorerformance.

Fig. 1 shows atomic structure of �-phase PVDF and �-phaseVDF respectively. The crystal structure for �-phase PVDF showow the dipoles are aligned parallel to each other to create surfaceharge separation under stress or compression. Due to the randomrientation of the dipoles in the �-phase, the dipoles cancel out eachther resulting in no net surface charges [4]. Untreated PVDF itselfannot have � phase without delicate mechanical stretching orlectrical poling processes. Additionally, due to its incompatibilityith the standard lithography process, many alternative fabrica-

ion methods such as screen printing and shadow mask process11] have been developed.

Our approach offers a practical option to overcome these lim-tations, and offers cost-effective batch process with high filmniformity and high resolution of polymer patterning towardsnhanced sensor performance using standard lithography fabri-ation. Further, we hypothesized that, without any mechanicaltretching or poling, a thinner film should exhibit higher piezoelec-ricity compared to a thicker film. The output voltage V producedy a piezoelectric sensor is given by [2]:

= F · g3n · d/A

here, F, g3n are the applied force, appropriate voltage coefficient; and A are the thickness and area of the film, respectively. Hence,

thicker film should have higher output voltage if the film exhibitsood �-phase structure. But a thinner film may have higher crys-allinity and hence higher voltage coefficient.

To verify the above, we fabricated two different thicknessesf PVDF films and characterized using Raman spectroscopy andTIR (Fourier Transform Infra Red) spectroscopy. Samples werepin coated on silicon wafer and the absorbance due to silicon was

Fig. 2. Raman spectroscopy measurements for the various PVDF–TrFE films used inthe present study, showing enhancement of beta phase on curing of films.

subtracted from a background spectrum. Raman spectroscopy wasdone using Renishaw InVia Raman spectroscope using green laser(514.4 nm) with exposure of 30 s.

Fig. 2 showing the Raman spectroscopy peaks for the variousPVDF film thicknesses are highly distinct. We notice a clear sharp-ening in the �-phase crystal phase (840 cm−1) on curing the 1 �mand 6 �m thick PVDF–TrFE films. Further, we notice that the �-phase crystals (800 cm−1) in the PVDF–TrFE films decreases or geteliminated as the film is heated and recystallized. This phenomenonhas also been reported elsewhere [12]. The intensity of the peakfrom PVDF absorbance (1430 cm−1) remains unaltered, however.Raman spectrum shows mainly the presence of the �-phase crys-tal domains throughout the films. Less noticeable, however, is thepresence of the �-phase crystal structures also. Fourier TransformInfra-Red (FTIR) absorption spectrum for the various PVDF filmsfabricated was also done in the present study (data not shown).The plots showed similar nature as that from Raman spectroscopy,confirming the presence of beta phase formation in the membrane.

3. Device fabrication

Two different designs of electrodes on PVDF were fabricated:(1) parallel type: electrodes on the same side, where the electrodecan be further patterned into dual and quadruple design; and (2)sandwich type: PVDF thin film sandwiched between the electrodes.The PVDF–TrFE lithographic based fabrication process for sandwichstructure is illustrated in Fig. 3. Starting with a silicon wafer, pho-toresist was patterned to define the bottom Al electrode (2000 Athick) using the lift-off process. PVDF–TrFE was spin coated on theAl-patterned wafer. Then another patterned photoresist was usedas a Reactive Ion Etch (RIE) mask to pattern the PVDF–TrFE thinfilm. The dry etch was performed under 100 sccm oxygen gas envi-ronment with 200 W RF power and 5 mT pressure. The PVDF–TrFEthin film etch rate was 150 nm/min. The photoresist mask wasetched simultaneously with an etch rate of 100 nm/min and dryetch was continued until the photoresist mask was etched fully.Top Cu electrode (2000 A thick) was patterned by wet etch and thephotoresist mask was removed by dry etch. Using this recipe, wefabricated single, dual, triple and quadruple membrane geometriesin the following bottom electrode sizes: 1 × 1 and 1.5 × 1.5 cm2.

Further, parallel electrode devices were also patterned using justthe top mask to pattern the electrodes on one side of the membrane.

Fig. 4 shows film thickness for PVDF–TrFE films that can beprepared versus spin speed. Three different solutions, having

T. Sharma et al. / Sensors and Actuators A 177 (2012) 87– 92 89

Pwt2a

phaipa

s0

F

Fig. 5. SEM images of PVDF–TrFE thin film on substrate. (a) 20 wt% PVDF–TrFEshowing poor adhesion under regular vacuum oven heat treatment; and (b) 8 wt%PVDF–TrFE, showing strong adhesion under vacuum oven heat treatment.

Fig. 3. Fabrication process for PVDF–TrFE devices.

VDF–TrFE concentrations of 8%, 15%, and 20% w/v in 2-butanone,ere prepared. Thickness decreased as spin speed increased. In

he present study, we used 8% w/v PVDF–TrFE copolymer with000 rpm for 30 s to get 1 �m thickness and 20% w/v PVDF–TrFEt 4000 rpm for 30 s to get 6 �m thickness.

Fig. 5(a) shows SEM images of 20% w/v PVDF–TrFE, showingoor adhesion on the silicon substrate under regular vacuum oveneat treatment and Fig. 5(b) 8% w/v PVDF–TrFE, showing strongdhesion under vacuum oven heat treatment. The film was curedn vacuum oven at 130 ◦C for 5 h for strong adhesion and near �-hase formation due to the residual stress between the thin filmnd the substrate.

Fabricated metal-PVDF–TrFE-metal (MIM) structured pressureensors (dual and quadruple PVDF–TrFE membrane) size of 0.25,.5, 1, and 1.5 cm are shown in Fig. 6.

ig. 4. Film thickness vs. spin speed with various weight/volume in MEK solution.

Fig. 6. Fabricated MIM structured pressure sensors of dual and quadruple mem-brane devices (bottom half) showing the device dimensions. Length ‘a’ was 1.0 or1.5 cm.

4. Experimental setup

Fig. 7 shows the schematic of the pressure sensor measure-ment in air chamber. The pressure on the PVDF sensor wasdetermined using commercial pressure sensor (Freescale Semicon-ductors, MPX2300DT1). Any possible leakages from the chamber

were prevented by the use of neoprene o-ring and vacuum grease.Air pressure and flow inside the chamber was manipulated usingthe multiple outlet valves.

Fig. 7. (a) Schematic showing chamber with air flow through the system; and (b)actual photograph of chamber in the testing setup.

9 d Actuators A 177 (2012) 87– 92

(avtp

bmwwdfstvak0swIpisdsnt

5

wFPweot0

qsa

fcisPtistttwbh

paPr

Fig. 8. Experimental results of output voltage for pressure sensor having dual

in the two plots was 2.3 for dual over the quadruple. This ratioof slope was as expected due to nearly double functionally active

-10

0

10

20

30

40

50

60

70

80

90

Pre

ssur

e (m

mH

g)

Vol

tage

(mV

)

Control Devices 1um PVDF Devices

010020030040050060070080090010001100120013001400150016001700180019002000

0 T. Sharma et al. / Sensors an

Air flow into the chamber was controlled using a flowmeterDwyer instruments) before the inlet and a multiple outlet valvefter the inlet. Flow rate combined with the number of open outletalves were used to generate air pressure inside the chamber inhe 0–300 mmHg range. This also happens to be the physiologicalressure range over which the PVDF sensor needs to be operational.

The PVDF device was placed at the center of the chamber. Theackside of the device was taped to the device holding plate. A com-ercial pressure sensor (Freescale Semiconductor, MPX2300DT1)as placed on the second plate. Both, the device and the sensorere so placed that the sensing element was perpendicular to theirection of the air flow and hence compression-induced signalrom the commercial sensor and the PVDF devices due to air pres-ure was recorded. The inlet was kept at constant air flow, whilehe exit of air from chamber was controlled using a multi-stagealve. The two electrodes of the PVDF devices were connected ton external charge amplifier. Charge amplifier setting were alwaysept on 40 dB gain (100× amplification), 100 nF feedback capacitor,.1–10 Hz bandpass. All data was recorded after 40 dB gain but wascaled to 0 dB before plotting. Outputs from the charge amplifierere connected to a USB-type (6009) Data Acquisition Kit (National

nstruments). The output was further filtered using a digital low-ass filter (set to 8 Hz) with Infinite impulse response filter set to

nverse Chebyshev filter of order three. The chamber plates wereealed using vacuum grease (Dow Corning) and o-ring for the PVDFevices. An actual picture of the chamber with the device in place ishown in Fig. 7. The electrode pads of the PVDF devices were con-ected to Cu tape through silver print (GC electronics). Wires werehen soldered to top of the Cu tape for external connectivity.

. Results and discussion

The responses upon intermittent external stress applicationsere immediate and very reliable, having very fast recovery time.

ig. 8 shows experimental results of output voltage for a dualVDF–TrFE film sensor (1 × 1 cm2 bottom electrode surface area)hen exposed to impulses of air pressure in the chamber. With

xposed to the identical pressure application, similar outputs werebserved though the charge amplifier. The two different MIM struc-ures on the same device showed a fast recovery time, 0.17 s and.23 s, respectively.

Such recovery times correspond to an efficient resolving fre-uency of up to 5 Hz. This happens to be in the physiologicalampling rates, as most of the physiological phenomenon occurst 1–3 Hz, corresponding to 72–180 beats per minute.

Fig. 9 shows the hold-and-release output response obtainedrom a single PVDF pressure sensor fabricated (in blue color; online)orresponding to the pressure the PVDF device was subjected to,nside the chamber. The response obtained from the PVDF deviceshow the typical dynamic response expected from piezoelectricVDF films. The pressure inside the chamber was measured usinghe commercial pressure sensor, indicated by the right vertical axisn Fig. 9. The negative voltage peak corresponds to the compres-ion of the membrane and the positive direction peak correspondso the relaxation of the membrane. The relaxation response fromhe PVDF film is a combined function of its spring constant andhe damping factor. Response from control devices is shown in red,hen exposed to similar pressures. Control devices were assem-

led in exactly the same manner as normal devices but did notave any PVDF membrane present.

For different magnitudes of pressure inside the chamber, the

eak voltage from the PVDF devices were obtained and plotteds shown in Fig. 10. The graph shows a linear response for theVDF based pressure sensors in physiologically relevant pressureange. Further, we notice that the slope is higher for 1 �m devices

PVDF–TrFE film, showing 26% recovery time mismatch between two membranesand 9% repeatability.

compared to 6 �m devices, indicating higher sensitivity from thin-ner devices. Results indicate that the fabricated pressure sensorshows near �-phase formation in the film even after standardlithography techniques.

Fig. 10a compares the dual membrane design with the quadru-ple membrane design for 1.5 × 1.5 cm2 devices. Only one of the topavailable electrodes was connected at a given time. The standarddeviation for all the data observed was low (<4%). The ratio of slope

403020100Time (s)

Fig. 9. Response from PVDF–TrFE devices on hold-and-release pressures inside theair chamber showing the true dynamic behavior of the piezoelectric PVDF films.

T. Sharma et al. / Sensors and Actu

1401201008060402020

40

60

80

100

120

140

160V

olta

ge (m

V)

Pressure (mmHg)

Dual Membrane Quadruple Membrane

300250200150100500

0

5

10

15

20

25

30

35

40

Vol

tage

(mV

)

Pressure (mmHg)

1μm PVDF Parallel Electrodes 1μm PVDF Sandwich Electrodes

280260240220200180160140120100806040

14161820222426283032343638

Vol

tage

(mV

)

Pressure (mmHg)

1 μm PVDF Film 6 μm PVDF Film

a

b

c

Fig. 10. Characterization of design variations for PVDF–TrFE devices (1.5 × 1.5 cm2).(a) High sensitivity of dual membranes comparing to quadruple membrane of 1 �mthick PVDF film, the ratio of the slope was 2.35; (b) higher sensitivity of sand-wp

sq

Peww

using PVDF–TrFE copolymer, Journal of Microelectromechanical Systems 17 (2)

ich devices (3.33 times) comparing to parallel electrode on devices; (c) sensingerformance of thin and thick films in the air chamber.

urface area for the dual membrane design compared to theuadruple membrane design.

This ratio of slope was also found to remain the same when theVDF film was not etched, indicating that it is the area between the

lectrodes which contributes towards the signal output. In otherords, there was no contribution from areas of the PVDF film whichere not covered by metal electrodes.

ators A 177 (2012) 87– 92 91

Fig. 10b compares the performance from PVDF devices with Par-allel electrodes vs. the Sandwich type electrodes. As expected, thesandwich type devices perform better (more than 3 times highersensitivity) compared to parallel electrodes. This is because thepiezoelectric coefficient for PVDF working in compression mode,d31, can be more than two times higher than the lateral stretchingmode, d33 [13].

Fig. 10c shows the improved performance obtained from a thin-ner film (1 �m) compared to a thicker (6 �m) PVDF film.

We observed nearly 1.7 times higher signal output from the1 �m PVDF device compared to 6 �m devices, indicative of thehigher piezoelectricity of thinner film. This is possible due to thehigher surface tension involved in the formation of a thinner filmduring spin coating, resulting in higher crystallinity in thinner filmscompared to the thicker films.

6. Conclusion

The low-cost, low-temperature standard lithography fabrica-tion technique is applicable to pattern biocompatible PVDF–TrFEcopolymer. The process is fully compatible with existing micro-machining fabrication processes without additional mechanicalstretching and electrical poling processes. Further, we found thatthe electrode-patterned surface area only contributes towardsthe output signal rather than the PVDF area. We achieved bet-ter �-phase formation in thin film (1 �m) PVDF–TrFE copolymercompared to the thick film (6 �m) without any electrical pol-ing or mechanical stretching. The demonstrated fast recoverytime (0.17 s), biocompatibility, and compact form factor show thegreat potential for pressure and flow direction measurements asimplantable biomedical devices.

Acknowledgments

This work was supported partially by National Science Founda-tion (grant number: 1128677), and the University of Texas HealthScience Center at Houston. We also appreciate MicroelectronicsResearch Center (MRC) and National Nanotechnology Infrastruc-ture Network (NNIN) staff for their continuing support. We thankDr. Ye Hu for help with the FTIR spectrum, and Dr. Jaclyn Wiggins-Camacho for help with the Raman spectroscopy.

References

[1] J.D. Sandham, et al., A randomized controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients, New England Journal of Medicine348 (1) (2003) 5–14.

[2] V.A. DePalma, et al., Investigation of three-surface properties of several met-als and their relation to blood compatibility, Journal of Biomedical MaterialsResearch 6 (4) (1972) 37–75.

[3] T.G. Vrachliotis, et al., Impact of unilateral common iliac vein occlusion on trap-ping efficacy of the greenfield filter: an in vitro study, Academic Radiology 8(6) (2001) 494–500.

[4] J. Ji, et al., An ultraminiature CMOS pressure sensor for a multiplexed car-diovascular catheter, IEEE Transactions on Electron Devices 39 (10) (1992)2260–2267.

[5] C. Hin-Leung, K.D. Wise, An ultraminiature solid-state pressure sensor for acardiovascular catheter, IEEE Transactions on Electron Devices 35 (12) (1988)2355–2362.

[6] J. Palmaz, Intravascular stents: tissue-stent interactions and design considera-tions, American Journal of Roentgenology 160 (3) (1993) 613–618.

[7] V. Parsonnet, et al., Thirty-one years of clinical experience with “nuclear-powered” pacemakers, Pacing and Clinical Electrophysiology 29 (2) (2006)195–200.

[8] L. Chunyan, et al., Flexible dome and bump shape piezoelectric tactile sensors

(2008) 334–341.[9] C. Li, et al., Dual-mode operation of flexible piezoelectric polymer diaphragm

for intracranial pressure measurement, Applied Physics Letters 96 (5) (2010)053502–053503.

9 d Actu

[

[

[

2 T. Sharma et al. / Sensors an

10] J. Sang-Soo, et al., A thin-film piezoelectric PVDF–TrFE based implantable pres-

sure sensor using lithographic patterning, IEEE 24th International Conferenceon Micro Electro Mechanical Systems (MEMS) (2011) 2011.

11] S. Kärki, J. Lekkala, A new method to measure heart rate with EMFi andPVDF materials, Journal of Medical Engineering & Technology 33 (7) (2009)551–558.

[

ators A 177 (2012) 87– 92

12] K. Tashiro, M. Kobayashi, Vibrational spectroscopic study of the ferroelectric

phase transition in vinylidene fluoride-trifluoroethylene copolymers: 1 Tem-perature dependence of the Raman spectra, Polymer 29 (3) (1988) 426–436.

13] T. Furukawa, T. Wang, in: T. Wang, J. Herbert, A. Glass (Eds.), Measurementsand Properties of Ferroelectric Polymers in the Applications of FerroelectricPolymers, Blackie, 1988, pp. 65–117.