This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

JOURNAL OF MICROELECTROMECHANICAL SYSTEMS 1

A Swallowable Smart Pill for Local Drug Delivery

Rosa Goffredo, Student Member, IEEE, Alessandro Pecora, Luca Maiolo, Andrea Ferrone,
Eugenio Guglielmelli, Senior Member, IEEE, and Dino Accoto, Member, IEEE

Abstract— In this paper, a smart pill for local drug delivery is its high patient acceptance and low cost [2]. Smart pills for
presented. The pill is designed to achieve local therapy on gas- drug delivery offer a number of significant opportunities for
trointestinal tissue thanks to a novel drug delivery system (DDS). pharmaceutical industries because they may be used in a wide
The DDS consists of a miniature electrolytic pump. This micro-
fabricated pump is ideal for medical applications because of its range of applications and enable therapies not possible with
low power consumption, compatibility with small size batteries, conventional means. For instance, they allow the treatment
integrability on miniaturized systems, and biocompatibility. The of locally active diseases along the gastrointestinal system,
actuation principle of the micropump relies on the electrolysis of such as gastroesophageal reflux (GERD), esophageal cancer
a water-based solution, which is separated from a drug reservoir or inflammatory bowel diseases (IBD), intestinal cancer and
by an elastic membrane. The electrolytically produced gases pres-
surize the electrolytic solution reservoir, causing the deflection of irritable bowel syndrome [3].
the elastic membrane. Such deflection, in turn, forces the drug In fact, therapies for those pathologies include biological
out of its reservoir through a nozzle. In order to measure and drugs that are very expensive and require large intravenous
monitor the membrane displacement, and therefore the volume infusions (e.g. infliximab, used in IBD, requires infusion
of drug ejected, a strain gauge sensor has been prepared using of 5 mg per kilogram of body weight from every
a conductive thermoplastic nanocomposite elastomer (CTPE).
The sensor is fixed on the deformable membrane. The CTPE 2 to 8 weeks [4]). The use of smart pills for local delivery
shows high sensitivity and allows to customize the resistance can drastically reduce the doses (∼50-60 μl are required)
of the device to obtain low power consumption. Thus, by and consequently the whole costs of the therapy. Swallowable
measuring the current though the electrolytic cell and monitoring drug delivery systems for local treatment bring also many
membrane deformation, the volume of the ejected drug can be advantages and opportunities for personalized therapies
controlled. [2015-0200]
(e.g. the region of interest can vary from patient to patient
Index Terms— Smart pill, drug delivery systems, miniature and for the same patient overtime), effectiveness (e.g. specific
electrolytic pump, conductive thermoplastic nanocomposite dose), better management of disease (e.g. site-specific drug
elastomer (CTPE) sensor, closed loop control.
delivery with minimal side-effects) reduced doses and
correspondingly lower costs for the health systems.
I. I NTRODUCTION The development of miniaturized fluidic systems and in
particular of (low-cost and biocompatible) micro pumps is
S MART pills are ingestible miniaturized electro-
mechanical devices [1], representing a point of con-
vergence between biomedical technology, medicine and the
functional to the improvement of the performance of drug
delivery systems (DDS).
pharma industry. Electronics, sensors and miniature robotic In general, the activation of a drug delivery system can
technology can give access, analyze and manipulate the body be triggered by different types of parameters, such as tem-
from the inside. In particular, smart pills for drug delivery are perature, chemical (i.e. pH) or electrical parameters (i.e.
an emerging technology; many different approaches to local impedance), that are medical indexes for the detection of
drug delivery have been proposed, including transcutaneous physiological conditions alterations. We chose to use on the
and implantable means. Anyhow, swallowable smart pills for capsule an electrical parameter, in particular an impedance
drug delivery are receiving increasing attention as the oral sensor, because biological tissues have different response
one is still the preferred route for drug administration, due to to different frequencies of stimulation and types of tissue
can be discriminated to their response. In addition, altered
Manuscript received July 14, 2015; revised January 24, 2016; accepted electrical responses could reveal pathological conditions [5].
January 26, 2016. Subject Editor C. Ahn.
R. Goffredo, E. Guglielmelli, and D. Accoto are with the Biomedical Furthermore, impedance measurement is less susceptible to
Robotics and Biomicrosystems Laboratory, Università Campus Bio-Medico external factors such as food or liquid ingestion than pH. Thus,
di Roma, Rome 00128, Italy (e-mail: [email protected]; activation of the drug delivery system according to the readings
[email protected]; [email protected]).
A. Pecora and L. Maiolo are with the Institute for Microelectronics of an impedance sensor, makes the smart pill appropriate for
and Microsystems, Consiglio Nazionale delle Ricerche, Rome 00133, Italy local treatment of site-specific active pathologies. Regardless,
(e-mail: [email protected]; [email protected]). the sensing system developed has been described by authors
A. Ferrone is with Institute for Microelectronics and Microsystems,
Consiglio Nazionale delle Ricerche, Rome 00133, Italy, and also with the in [6], so this paper will focus on the therapeutic functions of
Science Department, Università degli Studi Roma Tre, Rome 00146, Italy the capsule, i.e. the DDS. The design and fabrication of the
(e-mail: [email protected]). capsule is described, and the drug delivery system, comprising
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. an electrolytic microfabricated pump is presented and tested.
Digital Object Identifier 10.1109/JMEMS.2016.2524542 Furthermore, in order to enable the future design of a closed
1057-7157 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS

Fig. 3. Electrolytic pump (Reservoir volume 120 μl, PDMS membrane

Fig. 1. Capsule mock up and cross section. A. Drug Reservoir, B. PDMS
membrane C. Electrolyte reservoir, D. Electrodes E. Electronic board, area 40 mm2 thickness 60 μm , CTPE sensor area 10 mm2 ).
F. Battery (LIR1025, GMB Power), G. Electrodes for impedance measure-
ment H. Nozzle. Capsule dimensions are 21 × 14 mm (length × diameter).
management at microscopic volumes, that can be used as
active elements of smart pills. By their nature, pumps for
biomedical applications have to comply with stringent require-
ments in terms of reliability and flow-rate control [7].
In addition, swallowable drug delivery systems for gastroin-
testinal applications must be small in size, should operate
at low voltages (∼ 3V) and with low current intensities
(∼ 2 mA) [8]. Taking into account such requirements,
electrolytic pumps are a good solution to be integrated on a
swallowable capsule for targeted therapies. Electrolysis-based
Fig. 2. Schematic of the drug delivery system. The two reservoirs are pumps exploit gas production to deform a membrane and
separated by an elastic membrane (A). When current flows through the
electrodes, electrolytic reaction starts (B), the membrane deforms and drug is displace a fluid in contact with it [9].
released through the nozzle (C). The membrane is sensorized with a custom The electrolytic reaction is activated by a low voltage,
made strain sensor. compatible with microelectronics powering systems and with
small size batteries, which can be mounted on board.
loop control on the drug delivery, the electrolytic pump is The pump (fig. 3) contains 120 μl of non-toxic saline
sensorized with a novel strain sensor obtained by a conductive solution (NaCl in water, 0.5 M). The bottom of the reservoir
nanocomposite thermoplastic elastomer. Sensor properties and lodges two Pt electrodes (φ1 mm) that protrude in the solution
experimental tests on the systems are discussed. reservoir for 3 mm. Outside the reservoir the electrodes are
connected to a power supply system.
II. S YSTEM OVERVIEW The electrolytic reaction occurring when a voltage is applied
to the two inert electrodes is:
A. Fabrication
The capsule body (Fig. 1) is 3D printed in acrylic resin  4H+ + 4e− + O2 (g)
Anode : 2H2 O −→
(EX200 and S100 Visi Jet, HD3000 Projet, 3D Systems). Cathode : 2H2 O + 2e− −→  2OH− + H2 (g)
The total volume of the capsule is about 1300 mm3 . The
Net : 2H2 O(l) −→
←− O2 (g) + 2H2 (g)

drug delivery system comprises two reservoirs, respectively for
the drug and the electrolytic solution, separated by an elastic Overall, 1 mole of electrons is needed to generate 34 moles of
membrane. The two reservoirs occupy one lobe of the capsule gases (H2 and O2 ).
(Fig 1). The drug is delivered through a nozzle (φ 1 mm) on The electrolytic solution reservoir is closed by a PDMS
the frontal lobe of the capsule. The nozzle is normally closed membrane. The membrane is sensorized with a novel strain
by a pre-stretched paraffin membrane (Parafilm “M”, Pechey gauge sensor made of a conductive thermoplastic nanocom-
Plastic Packaging). posite elastomer (CTPE), in order to measure the overall
Two through holes on the membrane (φ 0.5 mm), placed deformations of the system (fig. 3).
on both sides of the nozzle, allow the drug to flow out of The area of the membrane separating the two reservoirs
the nozzle only when the paraffin membrane is deflected by is about 40 mm2 . The membrane is in PDMS (Sylgard 184,
the pressurized drug (Fig.2). This solution eliminates the use Dow Corning, Midland, MI). For fabricating the elastic mem-
of microvalves that are complex elements to integrate on the brane, PDMS is mixed with curing agent at 10:1 mass ratio.
dome of the capsule and could be damaged by the possible The mixture is degassed for 30 minutes, spun at 900 rpm
contact with the intestinal walls during the navigation. and cured at 60 °C in an oven for 30 minutes, according
to instructions reported in the manifacturer’s datasheet. The
B. Electrolytic Pump thickness of the membrane is about 60 μm. Spinning speed is
Bio-MEMS give the possibility to provide drug delivery determined experimentally in order to obtain a thin membrane
with microchannels, microvalves and micropumps for flow that can still be easily handled during cutting and mounting
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

GOFFREDO et al.: SWALLOWABLE SMART PILL FOR LOCAL DRUG DELIVERY 3

Fig. 5. Different response of a conductive thermoplastic material after the

Fig. 4. Cross section view of the drug delivery system. pre-treatment (Flandin curve) [10].

on the reservoir. The membrane is then cut and the CTPE An interesting property of the CTPE is that it can work in
sensor is fixed to the membrane, which can then be glued four different regimes (fig. 5) according to the pre-treatment to
to the electrolytic solution reservoir prefilled with the saline which the material is subjected. So the stripe of CTPE, used as
solution. a sensor, underwent a pre-strain procedure to force the device
The electrolytic pump is mounted in the frontal lobe of to work in the recoverable damage regime (regime III).
the capsule. The displacement of the membrane, due to gases This regime was selected to obtain a sensor with repeatable
production, forces the release of the drug (about 50 μl are characteristics and a customized value of resistance that can
required for the DDS) (fig. 4). guarantee low power consumption, necessary for the applica-
tion of the device in a DDS.
The procedure consisted of stretching the CTPE stripe to
C. Assembling
break the bonds inside the polymer and stabilize the resistance
The 3D printed capsule can be easily mechanically assem- to a percentage of the desired strain, thus obtaining a length
bled in few steps. on average three times the starting one and a resistance
Firstly, the drug delivery system is assembled: R0 ∼ 80k.
- The drug reservoir is filled from the bottom while the Electrical connections are provided at the edge of the stripe
dispensing hole on the drug reservoir is sealed. in order to read its resistance. During the experiments the
- The pre-filled electrolytic solution reservoir is fitted in variation of resistance was continuously measured with a
the drug reservoir that constitutes the frontal lobe of the multimeter (Hp1444a) and plotted in a Matlab Inc. interface.
capsule. CTPE has been chosen as the optimal candidate for con-
Then, the drug delivery system is mounted on the stant monitoring of the membrane deformations because it
remaining part of the capsule: shows a better behavior for low speed deformations, such
- the assembly (drug reservoir and electrolytic pump) is as those involved in electrolytic reactions (10 ÷ 20 μm/s),
locked by clamping on the central part of the cap- in comparison to other viscoelastic materials and nanocom-
sule a thin collar that prevents the electrolytic solution posites (e.g. PDMS). These materials, in fact, generally suffer
reservoir from sliding backward, while it cannot slide from a high drift, show very slow recovery times and cannot
forward because it mechanically fits the frontal lobe of track movements for low strain rates. Furthermore, a CTPE
the capsule. sensor can track deformations from 0.07% to 20%, while
- External electrodes are fixed on both sides of the central maintaining a good sensitivity to small movements and accept-
part of the capsule body. Finally, able hysteresis [11].
- The electronic board and the battery are put in place Preliminary strain tests on the conductive thermoplastic
in the central part of the capsule. nanocomposite have been carried out by cyclically stretching
- The capsule is closed by locking the second lobe on the the device along a single direction using a materials testing
rear part of the central body. machine (Instron 3365) at a speed of 100μm/sec at different
applied strain (5%, 10% and 20%). In such tests the number of
III. M ODELING AND T ESTING iterations is set to five. Considering the expected application,
this value corresponds to an overestimation of the number of
A. Strain Sensor Properties and Characterization actual membrane deformations that would occur in case of
The CTPE material was provided by Kraiburg TPE intermittent drug releases. The smallest hysteresis is shown
GmbH & Co: this material exhibits a density of 1.07 g/cm3, for 5% strain (fig. 6).
a hardness of 84 shA, resistance to traction of 10.9 N/mm2 Since the amplitude of the hysteresis cycle is propor-
and a wide working temperature range (−40÷110 °C). The tional to strain-rate, hysteresis can be neglected at sufficiently
strain sensor is cut from the CTPE material in a stripe low deformation speed as those involved in the electrolytic
(30 × 1 × 0.2 mm) with a value of resistance Rtpe ∼ 15 k. reaction.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS

Fig. 8. Membrane CAD and deflection model of the membrane due to the
applied pressure.

considering a rectangular membrane with sides 2a and 2b

Fig. 6. Hysteresis for the CTPE sensor obtained at different strain % (a<b).
(5%, 10%, 20%) at a speed of 100 μm/s. In the analytical model, a functional form of the deflected
shape is assumed and the total strain energy is minimized to
find the load-deflection behavior of the membrane [14].
According to the model, the total strain energy of the rectan-
gular membrane is obtained from the sum of the strain energy
of deformation and the elastic strain energy due to internal
tensile stress. In the case of large deflections (membrane
deflection w0  membrane thickness t) the strain energy due
to bending can be neglected. Usually, the Fourier expansion of
the true solution with two undetermined constants is chosen
for the functional form of the displacement of a point in the
membrane. Then the total potential energy of the membrane
is obtained from the work input into the membrane. Energy
is minimized resulting in a relationship between the load and
the deflection:
Fig. 7. Behavior of the sensor (blue curve) for a one cycle traction C2 (n, ν) t Ew30
C1 (n)σ w0 t
(green curve). P= + (1)
a2 a4
where P is the applied pressure, w0 is the center of deflection,
Fig. 7 shows the sensor response during a cyclic traction n is a/b, t is the thickness, E is the Young modulus, ν the
trial at 5% strain. The values of the resistances reported are in plane Poisson Ratio, σ the residual stress; the dimensionless
normalized in respect to the value R0 at rest. constants C1 (n),C2 (n, ν) are geometry dependent.
The sensor response curve tracks the deformation curve. At large deflection the cubic term dominates and Young
The increase of resistance is due to increasing spacing of the modulus determines the mechanical behavior.
conductive nanoparticles inside the polymeric matrix. Small C1 (n) and C2 (n, ν) are determined analytically to be [15]:
deviations depend on the speed of deformation (12μm/s) and
on the overall amount of strain. π 4 (1 + n 2 )
C1 (n) =
64 
B. Membrane Deflection π6 9 + 2n 2 + 9n 4
C2 (n, ν) =
32(1−ν 2) 256
1) Elastic Model for a Rectangular Membrane Loaded With 
a Uniform Pressure: The membrane deformations have been 4+n +n 2 +4n3 −3nν(1+n)2
−
modeled in order to evaluate the drug volume Vd that can be 2{81π 2 (1+n 2 )+128n +ν[128n −9π 2(1+n 2 )]}
delivered by the drug delivery system. (2)
Thanks to the slow deformation of the membrane, a quasi-
static model can be adopted to compute the deflection of the Considering the membrane geometry 2a=6 mm and
membrane. For the same reason, the pressure acting on the 2b = 9 mm and assuming ν = 0.5 for PDMS, C1 and C2
membrane can be assumed to be uniform over the surface. can be calculated as C1 = 2.21 and C2 = 1.94. E is estimated
The load deflection model is a known method for measuring to be 1.4 MPa considering load-deflection results previously
the elastic properties of a membrane. Tabata et al. [12] obtained by the authors [1]. The value of the modulus of Young
developed an analytical model for rectangular membrane calculated is confirmed by the data on thin PDMS membrane
that can approximately fit the membrane of the electrolytic in literature [16].
solution reservoir. Circular shape models [13] have also been The analytical solution was compared with FEM analysis by
considered but the rectangular membrane model has been Pan et al. [17], and confirmed by other authors [18] and they
selected because it better approximates the real membrane. found that the functional form of the analytical results similar
Figure 8 schematically shows the membrane by approximately to the FEM solution. Considered that, the load-deflection
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

GOFFREDO et al.: SWALLOWABLE SMART PILL FOR LOCAL DRUG DELIVERY 5

TABLE I
S IMULATIONS R ESULTS

Fig. 9. Displaced volume vs. pressure applied.

behavior of the membrane was modelled using the finite

elements method. This choice allows to properly consider the
membrane and the CTPE sensor together and to calculate the
deformation of the membrane together with the deformation of
the CTPE sensor. Furthermore, deformations can be calculated
considering the true membrane shape avoiding the introduction Fig. 10. Simulation of the membrane deformation.
of further approximations in the model that, by its nature,
produces an approximate solution.
2) Finite Element Method Analysis: FEM model was setup in order to verify the performance of the developed
generated into Comsol Multiphysics and the load-deflection system. The volumetric displacements estimated by integration
behavior was simulated. from the central displacements of the membrane are then
The membrane was clamped at its edge and uniformly correlated to the displacements measured on the setup.
loaded by a pressure P. Large deformation solver was adopted.
The parameters used in simulation were as follows.
C. Experimental Tests on the Membrane
Membrane surface: 40 mm2 ; thickness: 60μm; material:
PDMS (Young’s modulus of 1.4 MPa and Poisson’s ratio Considering the electrolytic solution (NaCl 0.5 M), by
of 0.49). approximating the mixture of O2 and H2 to an ideal gas,
The membrane acts as a supporting structure for the CTPE the amount of gas that at the pressure P produces the vol-
strain sensor. The parameters adopted for the CTPE sensor ume displacement Vd can be calculated. Considering simu-
were: surface area 10 mm2 thickness 0.2 mm, Poisson’s mod- lation results, for P = 0.34 bar such quantity is 24 μmol,
ulus of 0.49 and Young’s modulus 2.23 MPa. Such Young’s corresponding to 532 μl of gases in STP conditions.
modulus, was evaluated as the mean value calculated from Then, gas production was measured on the system during
a set of stress-strain measures made on a specimen of the the electrolytic reaction in the pump, to compare the per-
thermoplastic nanocomposite polymer. formance with the maximum volume of gas expected from
Mesh used consists of 23280 elements. The results in Table I simulation.
show the central deflection of the membrane in correspondence Gas production in the electrolytic reservoir can be estimated
of the load applied. Volumetric displacement was calculated using Faraday’s law:
by integration.
Q M
The volume displaced increases linearly with the mem- m=( · ) (3)
brane deflection. Figure 9 shows the volumetric deflection of F z
membrane for increasing values of pressure. where m is the mass of gases developed at the elec-
The maximum membrane displacement obtained is 2.16 mm trodes, Q is the electrical charge, F is Faraday’s constant
corresponding to about 53 μl that is a suitable value for drug (96500 C mol−1 ), Mz , is the mass to charge ratio of the
delivery application (fig. 10). electrolyte. For the considered reaction, 24 μmol of gases are
As shown further on in the experimental section, data from produced by 33 μmol of electrons, corresponding to a total
simulation have been compared with measurements on the charge Q = 3.2 C.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS

Fig. 13. Closed loop control.

Fig. 11. Current at the electrodes vs. time.

Fig. 14. Setup 3D CAD: it presents a central cavity (A) of the same shape
Fig. 12. Key steps for gas production measurement.
of the electrolytic solution reservoir with a hole in the bottom to allow the
connection to a tube for air insufflation (B). On the surface of the setup the
PDMS membrane is laid and the CTPE fixed (C).
Sincethe electrical charge Q is the time integral of current I
τ
(Q = 0 I (t) dt), the amount of gases produced can be As the volume of gas produced, Vgas , is calculated from
evaluated by monitoring the current through the electrodes. the measured current at the electrodes and the variation of
To this purpose, an interface was implemented in Labview resistance is correlated to the volumetric displacement of the
(National Instrument) running on a PC equipped with a membrane, the latter can be adopted in the control system of
DAQ NI USB-6009 ADC/DAC converter card. An analog- the drug delivery (fig. 13).
to-digital line recorded the current through the pump when a Furthermore, the integration of the strain sensor, that moni-
constant 3V voltage was applied to the electrodes. Current tors the membrane deflections, makes the drug delivery system
flowing through the electrolytic cell is influenced by the appropriate for the dispensing of different kind of liquid drugs.
concentration of ions and increases over time (Fig.11). In fact, the measure is not influenced, for example, by the
The amount of gas produced in the pump, Vgas , is propor- fluid viscosity as will happen with a control based on a direct
tional to the electrical charge Q according to (3) and can be pressure measurement inside the capsule.
calculated following the logical flow diagram shown in Fig 12. The closed-loop control implemented is fundament for the
The number of moles of electrons produced in 120 sec, correct functioning of the smart pill as it is supposed to be
ne-, is ne-=Q/F=5.5 μmol, corresponding to 7.3 μmol of gas able to reject all possible failures during navigation. Possible
(164 μl of gas in STP conditions). malfunctions could be caused by the external pressure exerted
The relation between the current measured at the electrodes by gastric fluids on the capsule and in particular on the nozzle.
and the flow rate of the released drug is shown in eq. (4) as The monitoring of the membrane deformations permits to
a function of the time t: identify any cases of nozzle plugging. The measure of the

VST P τ current at the electrodes allows a double check for cases when
Q (t) = I (t) dt (4)
z Ft 0 the gas is produced in the electrolytic pump but no deformation
considering the flow diagram in fig 12, the volume of of the membrane is read because the nozzle is plugged by
gas produced is expressed as Vgas = n gas · VST external forces or cases when the membrane breaks and no
 P where, τ
I (t )dt deflection is detected by the CTPE sensor while the electrolytic
VST P = 22.414l and the moles of gas n gas = 0 z F are reaction is active.
calculated from the Faraday’s law (eq. 3), taking into account
that the electrical charge Q is the time integral of current I .
E. Experimental Tests on the Sensor and
So the flow rate calculated is 1,36 μl/s.
Electrolytic Pump Sensorization
In order to compare simulation results with real membrane
D. Membrane Sensorization deflections and the corresponding displaced volumes a setup
Sensorization of the membrane allows to measure the over- was fabricated. CTPE sensor resistances were measured in
all deformations of the membrane, as needed to achieve a order to obtain a correspondence between the value of resis-
closed-loop control of the amount of drug released. tance measured by the sensor and volumetric displacement
In fact, once known the information about the sensor resis- related to the amount of drug to deliver.
tance variations and the current increase due to the electrolytic The dedicated setup (fig. 14) was 3D printed in order to
reaction, the drug release can be controlled. have a stable structure for repeated tests. This setup permitted
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

GOFFREDO et al.: SWALLOWABLE SMART PILL FOR LOCAL DRUG DELIVERY 7

Fig. 17. Resistance Vs. applied load.

Fig. 15. Experimental test: the sensorized membrane is inflated from a
bottom inlet (A), the sensor response is read by a multimeter (B). Membrane
deformation is monitored from the top with a microscope (C).

Fig. 18. Resistance Vs. displaced volume.

Fig. 16. Load-deflection behavior.
Figure 17 shows the sensor response at increasing applied
to collect data about the displacement of the membrane and the pressures.
sensor response at different pressure’s values till the maximum Once the displacement of the center of the membrane is
value supposed to be produced during the electrolytic reaction. measured with the microscope, the corresponding volumetric
Those data can be used to implement the control and deflection can be retrieved from the simulations results.
have been verified on the drug delivery system during the The sensor response at increasing displaced volume is
electrolytic reaction. shown in fig. 18.
The setup presents a central cavity of the same shape of Data of CTPE normalized resistance R/R0 , corresponding to
the electrolytic solution reservoir with a hole in the bottom to increasing volume of gas production and so to drug delivered,
allow the connection to a tube for air insufflation. are reported in Table II. Such values can be used in the control
On the surface of the setup the PDMS membrane is laid and system.
the CTPE fixed on it with electrical connections at the edges The capability of the CTPE sensor to monitor small
in order to monitor resistance variations with a multimeter deformations and to detect rapid variations, as discussed
(Hp 1444A). The membrane is squeezed by the upper part in Sec. III. A, is confirmed by the experimental setup.
of the setup to ensure the chamber is leak tight. A hole on Figure 19 shows the sensor response to a continuous
this part in correspondence of the chamber in the bottom increase of pressure from 0.02 bar to 0.34 bar until the
side allows monitoring of the membrane deformation with a membrane is deflated: the CTPE maintains a good sensitivity
miscoscope (Superyes 3.5). The initial membrane deflection is to small displacements due to small pressure increases.
retrieved from the frame recorded with the microscope. Deflections of the membrane and sensor response have also
The membrane is then pressurized incrementally, from been measured on the pump during the electrolytic reaction
0.02 bar to 0.34 bar. Air is inflated from the bottom inlet and the volume of the drug delivered controlled. The pump was
with a fluid dispenser (Techcon Systems TS255) (fig. 15). powered at 3 V for about 50s. At the beginning of the reaction,
Load-deflection data are plotted in figure 16. the increase of the resistance is low because initial deformation
As the displacement increased with pressure also the is small, then, there is a rapid increase of resistance. The
resistance of the strain sensor increased. maximum value of the resistance measured at the sensor has
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS

TABLE II
S ENSOR R ESPONSE AT D IFFERENT V OLUMETRIC D ISPLACEMENTS

Fig. 20. Resistance vs. displaced volume during the electrolytic reaction.

IV. D ISCUSSION AND C ONCLUSIONS

In this work, a swallowable smart pill for local drug delivery
has been proposed.
The smart pill includes a drug delivery system feasible
for oral administration thanks to its small dimensions, low
power consumption (electrolytic pump, CTPE sensor) and
biocompatibility, assured by a proper selection of materials
(e.g. PDMS membrane in contact with drug; saline electrolytic
solution, Pt electrodes).
Drug is pumped by a custom made micro electrolytic pump.
The integrated electrolytic pump can deliver a useful amount
of drug ( >50μl) in about 1 min. Furthermore, the electrolytic
pump is equipped with a novel strain gauge sensor in order to
monitor membrane deformation due to gases production in the
electrolytic pump correlating it to a measured resistance. The
measure of the deformation of the membrane is functional to
the control of the drug released.
The strain sensor implemented is a conductive thermoplastic
nanocomposite elastomer that can be easily integrated on the
pump membrane. Results of the experimental tests showed
that CTPE can successfully be applied a strain sensor on
Fig. 19. Sensor response during pressure increase (0.02- 0.34 bar) and the electrolytic reservoir because of its high sensitivity to
deflation. any, even small, membrane displacement. Furthermore, a great
advantage given by the use of a CTPE sensor for this appli-
been R/R0 = 1.1. A volume Vd of 45 μl can be delivered in cation is the possibility to achieve low power consumption.
correspondence of the measured resistance (Fig. 20). In fact, as the sensor resistance can be customized choosing the
During the trial, the deflection was monitored with the regime in which the sensor works through the pre-treatment,
microscope. a high value of resistance (≤ 1 M) can be obtained. That is
A displacement of 1.8 mm is reached, that corresponds to very important considering that the smart pill is powered by
the expected volume, proving the accuracy of the measure. batteries on-board. In fact, all the components and solutions on
Therefore the gas produced by the electrolytic reaction gen- the pill have been designed to achieve low power consumption
erates a pressure on the membrane that can be estimated to (e.g. choice of the microcontroller, activation time of sensors
be of about 0.2 bar comparing displacement results with those and microcontroller, electrolytic pump for drug delivery).
obtained from simulation. The sensor response has been correlated to the volumes
Finally, as CTPE resistance can be significantly influenced of drug delivered. A dataset of expected volumes of drug
by changes in temperature [8], this value was monitored with released has been collected from simulations considering the
a thermocouple J during the electrolytic reaction. No appre- experimentally determined mechanical parameters of the
ciable changes in temperature (the trial was conducted at system and the pressure increments expected from the
room temperature) were measured so alteration of the CTPE electrolytic reaction. The drug delivery system is controlled
performances are excluded during the system activation on the measuring the current through the electrodes during the
capsule. Also, since electrolysis occurs at constant tempera- electrolytic reaction and so calculating the flow rate
ture, the performance of the sensor is not affected by thermal and monitoring the deformation of the electrolytic pump
effects during drug delivery. membrane with the CTPE strain sensor.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

GOFFREDO et al.: SWALLOWABLE SMART PILL FOR LOCAL DRUG DELIVERY 9

R EFERENCES Alessandro Pecora was born in Rome in 1962.

He received the Master’s degree in physics from the
[1] R. Goffredo, A. Ferrone, L. Maiolo, A. Pecora, and D. Accoto, University of Rome La Sapienza, in 1990, with a
“A miniaturized electrolytic pump sensorized with a strain gauge based dissertation on solid-surface electron spectroscopy.
on thermoplastic nanocomposite for drug delivery systems,” in Proc. Since 1990, he has been a Research Scientist with
EMBC, Milan, Italy, Aug. 2015, pp. 3205–3208. the Institute for Microelectronics and Microsystems,
[2] R. Goffredo, D. Accoto, and E. Guglielmelli, “Swallowable smart pills where he is responsible for the activity on devices for
for local drug delivery: Present status and future perspectives,” Expert large area electronics with specific focus on flexible
Rev. Med. Devices, vol. 12, no. 5, pp. 585–599, 2015. electronics and smart sensors.
[3] J. L. Gonzalez-Guillaumin, D. C. Sadowski, K. V. I. S. Kaler, and
M. P. Mintchev, “Ingestible capsule for impedance and pH monitor-
ing in the esophagus,” IEEE Trans. Biomed. Eng., vol. 54, no. 12,
pp. 2231–2236, Dec. 2007.
[4] J. F. Colombel, “Infliximab, azathioprine, or combination therapy Luca Maiolo received the Master’s degree in
for Crohn’s disease,” New England J. Med., vol. 362, no. 15, physics and the Ph.D. degree from the Università
pp. 1383–1395, 2010. degli studi di RomaTre, in 2003 and 2008, respec-
[5] S. Laufer, A. Ivorra, V. E. Reuter, B. Rubinsky, and S. B. Solomon, tively. In 2009, he joined the Robotics, Brain, and
“Electrical impedance characterization of normal and cancerous human Cognitive Sciences Department, Italian Institute of
hepatic tissue,” Physiol. Meas., vol. 31, no. 7, pp. 995–1009, Jul. 2010. Technology, Genoa, as a Postdoctoral Fellow, where
[6] R. Goffredo, D. Accoto, M. Santonico, G. Pennazza, and he developed and tested tactile flexible sensors for
E. Guglielmelli, “A smart pill for drug delivery with sensing humanoid robot. He is currently a Researcher with
capabilities,” in Proc. EMBC, Milan, Italy, Aug. 2015, the Institute for Microelectronics and Microsystems,
pp. 1361–1364. National Research Council. His activity is mainly
[7] J. M. Portilla and U. Kim, “Low-power, self-contained, reciprocat- focused on fabrication and characterization of elec-
ing micropump through electrolysis and catalyst-driven recombination tronic circuits, and smart sensors integrated on stretchable and flexible
toward drug delivery applications,” in Proc. MEMS, Taipei, Taiwan, substrates.
Jan. 2013, pp. 1077–1080.
[8] C. G. Cameron and M. S. Freund, “Electrolytic actuators: Alternative,
high-performance, material-based devices,” in Proc. Nat. Acad. Sci.,
vol. 99, no. 12, pp. 7827–7831, 2002. Andrea Ferrone received the Master’s degree in
[9] D. Accoto, V. Mattioli, P. Valdastri, A. Menciassi, and P. Dario, electronics engineering for industry and innovation
“A miniaturized drug-delivery system for intra-corporal use,” in Proc. from Università Roma Tre in 2014. He is cur-
8th Italian Conf. Sens. Microsyst., Trento, Italy, Feb. 2003, pp. 12–14. rently pursuing the Ph.D. degree with the Institute
[10] L. Flandin, A. Hiltner, and E. Baer, “Interrelationships between electrical for Microelectronics and Microsystems, National
and mechanical properties of a carbon black-filled ethylene–octene Research Council. His main activities focus on fab-
elastomer,” Polymer, vol. 42, no. 2, pp. 827–838, 2001. rication, characterization, and modeling of electronic
[11] A. Pecora et al., “Strain gauge sensors based on thermoplastic nanocom- and microelectronic sensors with nanocomposite
posite for monitoring inflatable structures,” in Proc. IEEE Metrol. materials.
Aerosp. (MetroAeroSpace), May 2014, pp. 84–88.
[12] O. Tabata, K. Kawahata, S. Sugiyama, and I. Igarashi, “Mechanical
property measurements of thin films using load-deflection of compos-
ite rectangular membranes,” Sens. Actuators, vol. 20, pp. 135–141,
Nov. 1989. Eugenio Guglielmelli (S’92–M’95–SM’11)
[13] O. Brand, G. K. Fedder, C. Hierold, and J. G. Korvink, Relia- received the Laurea degree in electronics
bility of MEMS: Testing of Materials and Devices, O. Tabata and engineering and the Ph.D. degree in biomedical
T. Tsuchiya, Eds. Hoboken, NJ, USA: Wiley, 2013. robotics from the University of Pisa, Italy, in 1991
[14] D. Maier-Schneider, J. Maibach, and E. Obermeier, “A new ana- and 1995, respectively. He is currently a Full
lytical solution for the load-deflection of square membranes,” Professor of Bioengineering with the Università
J. Microelectromech. Syst., vol. 4, no. 4, pp. 238–241, Dec. 1995. Campus Bio-Medico di Roma, Rome, Italy, where
[15] J. G. Korvink and O. Paul, MEMS: A Practical Guide to Design, he serves as the Head of the Laboratory of
Analysis, and Applications. New York, NY, USA: Springer, 2010. Biomedical Robotics and Biomicrosystems, which
[16] M. Liu, J. Sun, Y. Sun, C. Bock, and Q. Chen, “Thickness- he founded in 2004. He is a Principal Investigator
dependent mechanical properties of polydimethylsiloxane membranes,” and a Partner with several national and international
J. Micromech. Microeng., vol. 19, no. 3, p. 035028, 2009. projects in the area of biomedical robotics. His main current research
[17] J. Y. Pan, P. Lin, F. Maseeh, and S. D. Senturia, “Verification of interests are in the fields of human-centered robotics, biomechatronic
FEM analysis of load-deflection methods for measuring mechanical design, and biomorphic control of robotic systems and in their application
properties of thin films,” in IEEE Solid-State Sens. Actuator Workshop, to robot-mediated motor therapy, assistive robotics, and neurorobotics.
4th Tech. Dig., Jun. 1990, pp. 70–73. He served as an Associate Editor of the IEEE Robotics and Automation
[18] S. P. Timoshenko and S. Woinowsky-Krieger, Theory of Plates and Magazine, and the IEEE T RANSACTIONS ON ROBOTICS . He also serves as
Shells. New York, NY, USA: McGraw-Hill, 1959. the Editor-in-Chief of the Series on Biosystems and Biorobotics (Springer).

Dino Accoto (M’06) received the M.S.M.E.

(cum laude) degree in mechanical engineering from
the University of Pisa in 1998, and the Ph.D.
Rosa Goffredo (S’13) received the M.S. degree
(cum laude) degree in biomedical robotics from
in biomedical engineering and the Ph.D. degree in
Scuola Sant’Anna in 2002. From 2003 to 2007, he
biomedical engineering from the Università Campus
was an Assistant Professor of Biomedical Engineer-
Bio-Medico di Roma, Rome, Italy, in 2011 and
ing with Scuola Sant’Anna. Since 2004, he has been
2015, respectively.
with the Biomedical Robotics and Biomicrosys-
Her research interests include the development
tems Laboratory, Università Campus Bio-Medico
of electromechanical systems for biorobotics and
di Roma, Rome, Italy, where he is an Associate
biomedicine, microfluidics, therapeutic systems, and
Professor of Bioengineering. His current research
biosensors.
interest is the development of robotic and mechatronic systems for assistance,
rehabilitation, and surgery using multiscale and multiphysics approaches.