ABSTRACT 
 
 
 
 
Title of Dissertation: DEVELOPMENT OF AN INTEGRATED 
CAPSULE SYSTEM FOR 
GASTROINTESTINAL-TARGETED 
BIOSENSING  
  
 George Banis, Doctor of Philosophy, 2019 
  
Dissertation Directed By: Professor Reza Ghodssi, Department of 
Electrical and Computer Engineering, Institute 
for Systems Research 
 
 
Non-invasive microsystems are emerging as a means to address diagnostics 
challenges in healthcare due to the potential to retrieve information at the source and 
in a personalized approach. The gastrointestinal (GI) tract is a hub of information that 
alters in composition during both homeostatic and pathological conditions, and often 
manifests as varying biochemical concentrations in cell and tissue-sourced secretions. 
Thus, innovative strategies to sample molecular information from these secretions 
would be of significant benefit to physicians in establishing an appropriate prognosis. 
This dissertation describes the development of a film-based capacitive sensing strategy 
and subsequent integration into a capsule-based microsystem that is designed to travel 
through the GI tract upon ingestion until it passes through the stomach, where it is 
designed to measure model analytes in duodenal secretions. Subsequently, the 
measurements are processed into signals for wireless transmission, enabling external 
analysis for potential clinical utility.  
  
To achieve a system that can be safely ingested by patients, design features must 
be implemented that follow previously established standards in device requirements 
such as geometry and biocompatibility. In this work, I aided in the design, integration, 
and characterization of a capsule-embedded sensing system using commercial off-the-
shelf components that interface capacitive transducers (range: 0.8-220 pF; sensitivity: 
7.3x10-3) with a smart phone via Bluetooth Low Energy (2.4 GHz). The transducers 
are designed to measure the change in dielectric constant of interfacing media, which 
transitions when specific environmental (pH) characteristics are met. The system, 
including the power supply, are manufactured on a printed circuit board and packaged 
within a 3D-printed capsule structure (13 mm x 35 mm) that maintains dimensions of 
other clinically utilized ingestible capsule devices. The system is cost effective, user-
friendly, biocompatible, and can serve as a highly customizable platform for measuring 
a variety of desired targets. 
Secretions from various GI organs can be distinguished by pH, as is 
demonstrated in the pharmaceutical industry via enteric coatings that dissolve in target 
pH ranges but maintain structural stability in others. I employed such coatings for 
protecting our system until targeting the pH, and therefore GI region, of interest for 
sampling. Once dissolved, microfluidic inlets allow access for the media to interface 
with the sensors. I studied coatings that respond to both acidic (<pH 3) and neutral 
conditions (>pH 6), as well as pH sequences via hierarchical coatings. Because the 
target analytes react with naturally occurring substrates, I investigated label-free 
sensing of model enzymes such as pancreatic trypsin (20-40 ?M) and lipase (10 ?M-1 
mM), as well as bile salts (0.07-7 %w/v) as a model emulsifier, using films composed 
  
of biomaterials, including gelatin and stearin. To integrate these materials with the 
desired microsystem, I investigated various film deposition and modification strategies. 
Studies performed with our platform suggest the potential for the ability to sample the 
target fluid, as well as sense the analyte of interest in different concentrations by 
comparing the rate of capacitance change upon fluid entry compared to uncoated 
controls. Using this system, I characterized its potential for utility as a non-invasive 
platform for targeting multiple GI regions and detecting sensor-compatible biomarkers. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
DEVELOPMENT OF AN INTEGRATED CAPSULE SYSTEM FOR 
GASTROINTESTINAL-TARGETED BIOSENSING  
 
 
 
by 
 
 
George Efstratios Banis 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the 
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2019 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Professor Reza Ghodssi, Chair/Advisor 
Professor William Bentley 
Professor Ian White 
Professor Katharina Maisel 
Professor Peter Kofinas 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
George Efstratios Banis 
2019 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
Dedication 
To my parents, 
whose boundless support enabled my success, 
 
to Christina Angela, 
my love and best friend, 
 
and 
 
to Maraki, Ted, and Panagiotis Theodoros Adamos. 
ii 
 
 
 
Acknowledgements 
 First and foremost, I would like to thank my advisor Professor Reza Ghodssi 
for his guidance and support during the course of this work. I would also like to thank 
the members of the Ph.D. dissertation committee, Professor William Bentley, Professor 
Ian White, Professor Katharina Maisel, and Professor Peter Kofinas, for their helpful 
discussions leading to the completion of this work. 
 There a several individuals with whom this work would not been possible. I 
acknowledge Dr. Luke Beardslee for his engineering, scientific, and medical expertise 
and skills that both challenged me and led to the development of various technologies 
critical to this work. I am also grateful to other members of MEMS Sensors and 
Actuators Laboratory (MSAL), most notably Justin Stine and Rajendra Mayavan 
Sathyam, whose expertise significantly aided the progression of the system electronics. 
The success of this project developed further through extensive feedback and support 
from additional members of MSAL, including Dr. Sangwook Chu, Ryan Huiszoon, 
Ashley Chapin, Dr. Pradeep Rajasekaran, and Dr. Sanwei Liu. 
My experience in MSAL was initially shaped by my excellent mentors and 
members of the MiND project, Professor Hadar Ben-Yoav, Sheryl Chocron, and Dr. 
Thomas Winkler, who introduced me to the lab culture and set the standards for how 
to enhance my research and writing skills. My gratitude extends further to other 
members of the MiND project, specifically Dr. Gregory Payne, Dr. Eunkyoung Kim, 
Dr. Mijeong Kang, Dr. Deanna Kelly, Christopher Kitchen, and Stephanie Feldman. I 
also want to thank MSAL alumni Dr. Hyun Jung, Dr. Sowmya Subramanian, Dr. 
Faheng Zhang, and Dr. Konstantinos Gerasopoulos for their support, technical 
iii 
 
 
 
feedback, and overall useful discussions. I would also like to give mention to the 
undergraduate students involved with my projects in MSAL, including Patricia Barton, 
Florence Stevenson, Sonia Arya, Tony Feric, Charles Perera, Aditya Nilacantan, and 
Prateek Sayyaparaju. 
I would like to acknowledge the support and training I received in aspects of 
device fabrication, specifically staff members of the Maryland NanoCenter including 
those in the FabLab (Mr. John Abrahams, Mr. Thomas Loughran, Mr. Jonathan 
Hummel, and Mr. Mark Lecates), AIMLab (Dr. Wen-An Chiou, Dr. Sz-Chian Liou), 
and Terrapin Works. Moreover, I am grateful to the administrative staff in ISR and 
BioE, especially Vicci Barret, Regina King, Carla Scarbor, Dawn Wheeler, Dr. Tracy 
Chung, and Bill Churma, for their responsiveness and willingness to put up with my 
administrative concerns and request. Additionally, this work was only achieved 
through the funding from the National Science Foundation (NSF) under program 
manager Shubra Gangopadhyay. 
Finally, I would like to thank my family and friends. My parents sacrificed more 
than they needed to for me to succeed, and are my primary source of unyielding 
encouragement and support along with my sister Maria and her husband Ted. I have 
been very fortunate to have the support of some of my closest friends here in the BIOE 
department, namely Leopoldo Torres, Imaly Nanayakkara, Anjana Jeyaram, and 
Kelsey Gray, without whom fun and emotional support would have been significantly 
more difficult to achieve. Lastly, I want to thank my partner Christina, for her constant 
love and being my foundation for what matters most to me in life. 
iv 
 
 
 
Table of Contents 
 
Dedication ..................................................................................................................... ii 
Acknowledgements ...................................................................................................... iii 
Table of Contents .......................................................................................................... v 
List of Tables ............................................................................................................. viii 
List of Figures .............................................................................................................. ix 
List of Abbreviations ................................................................................................. xiv 
1. Chapter 1: Introduction ......................................................................................... 1 
1.1 Background and Motivation ............................................................................... 1 
1.2 Summary of Accomplishments ........................................................................... 3 
1.2.1 Investigation of Biomaterial Film-Based Electrical Transduction Systems 
for Measuring Enzymatic Activity........................................................................ 4 
1.2.2 Investigation of Embedded Packaging Materials for Gastrointestinal 
Targeting ............................................................................................................... 5 
1.2.3 Development of Ingestible Integrated System for Targeted Gastrointestinal 
Sampling and Enzyme Sensing ............................................................................. 6 
1.3 Literature Review................................................................................................ 7 
1.3.1 Electrical Biosensors .................................................................................... 7 
1.3.2 Exocrine Pancreatic Health and Diagnostics ............................................. 13 
1.3.3 Pancreatic Enzyme Activity and Other Gastrointestinal Fluid Contents ... 17 
1.3.4 Integrated In situ Capsule Technology ...................................................... 22 
1.4 Structure of Dissertation ................................................................................... 28 
2. Chapter 2: Electrical Transduction of Pancreatic Trypsin Activity .................... 29 
2.1 Pancreatic Enzyme Sensing .............................................................................. 29 
2.2 System Design and Experimental Setup ........................................................... 33 
2.2.1 IDE Sensor Fabrication and Film Deposition ............................................ 33 
2.2.2 Enzyme Solution Preparation .................................................................... 35 
2.2.3 Impedance Sensing .................................................................................... 35 
2.2.4 QCM Sensing Methodology ...................................................................... 37 
2.2.5 Image/Data Analysis .................................................................................. 37 
2.3 Results and Discussion ..................................................................................... 38 
2.3.1 Film Structure and Morphology................................................................. 38 
2.3.2 Impedance Sensing of Trypsin ................................................................... 41 
2.3.3 Impedance Response with Enzyme Mixture.............................................. 44 
2.3.4 QCM Results .............................................................................................. 48 
2.4 Film Reactions to Varying Enzyme Concentrations ......................................... 50 
2.5 System Modeling for Capacitive Sensing ......................................................... 53 
v 
 
 
 
2.5.1 Electrical Behavior of Biomaterial Films .................................................. 53 
2.5.3 Sensor Characterization with Varying Film Thickness ............................. 55 
2.6 Conclusions ....................................................................................................... 62 
3. Chapter 3: Packaging and System Miniaturization for Gastrointestinal 
Navigation ................................................................................................................... 64 
3.1 Capsules and Gastrointestinal Sampling ........................................................... 64 
3.2 System Description ........................................................................................... 70 
3.2.1 Sensors and External System Electronics .................................................. 70 
3.2.2 Capsule Package Design and Materials ..................................................... 72 
3.2.3 System Miniaturization and Assembly ...................................................... 74 
3.2.4 Experimental Procedures for pH Sequences .............................................. 76 
3.3 Experimental Results ........................................................................................ 78 
3.3.1 Sensor Responses ....................................................................................... 78 
3.3.2 Coating Variety and Combinations ............................................................ 79 
3.3.3 Assembled Capsule Coating and Sensor Responses .................................. 83 
3.4 Mechanical Testing ........................................................................................... 85 
4. Chapter 4:  Biochemical Capacitive Sensing with Triglyceride Films and Module 
Integration into Capsule System ................................................................................. 89 
4.1 Capsule-Based System Integration ................................................................... 89 
4.2 Pancreatic Lipase Sensing Strategies ................................................................ 90 
4.3 System Setup ..................................................................................................... 93 
4.3.1 Film Deposition ......................................................................................... 93 
4.3.2 Sensing Characterization ........................................................................... 96 
4.3.3 SEM Analysis for Film Morphology ......................................................... 97 
4.3.4 pH-Dependent Sampling and Sensing ....................................................... 98 
4.4 Results and Discussion ..................................................................................... 98 
4.5 Conclusion ...................................................................................................... 112 
5. Chapter 5:  Concluding Remarks ...................................................................... 114 
Highlights .............................................................................................................. 114 
5.1 Summary ......................................................................................................... 115 
5.2 Future Outlook ................................................................................................ 120 
5.2.1 Gastrointestinal-Simulating Biochemical and Biomechanical Conditions
........................................................................................................................... 121 
5.2.2 Anchoring Capsules ................................................................................. 124 
5.2.3 Feedback-Triggered Therapeutic Release ................................................ 126 
5.3 Conclusion ...................................................................................................... 128 
6. Appendices ........................................................................................................ 130 
6.1 Appendix A: Masks Used ............................................................................... 130 
A.1 Microfluidic Impedance Sensing ............................................................... 130 
A.2 First Capsule Mask for Capacitive Sensing ............................................... 131 
vi 
 
 
 
A.3 Sensors used for SPA ................................................................................. 132 
6.2 Appendix B: COMSOL Simulation and Results ............................................ 133 
6.4 Appendix C: MATLAB Analysis Code and GUI ........................................... 135 
7. Bibliography ..................................................................................................... 152 
 
 
 
 
 
 
 
 
vii 
 
 
 
List of Tables 
 
Table 1-1 Major gastrointestinal enzymes, organized by source and excluding the 
salivary glands and large intestine, and some of their respective associated health 
conditions. ................................................................................................................... 20 
Table 1-2 Integrated capsule systems. From left to right: PillCam COLON 2, GI gas 
sensor, and GI bleeding monitor. ................................................................................ 24 
Table 2-1 Ratio of spin-coated film remaining after various enzyme conditions, 
analyzed with a contact profiler (n = 2). ..................................................................... 39 
Table 2-2 Parameters used for capacitance analysis of gelatin films. ........................ 59 
Table 3-1 Various GI enzymes and their pH for optimal activity [16]. ...................... 68 
Table 3-2 System Specifications ................................................................................. 76 
Table 5-1 Summary of system parameters for the capsule device with reference to 
their respective chapters for more detail. .................................................................. 117 
Table 6-1 Parameters used for the COMSOL simulation. ........................................ 133 
 
viii 
 
 
 
List of Figures 
Figure 1-1 Variety of biosensing mechanisms [29]. ..................................................... 8 
Figure 1-2 Randles-Ershler equivalent circuit models for (a) Faradaic and (b) non-
Faradaic EIS. (c) Nyquist spectra for Faradaic EIS for a: ZW ? and Ret ? controlled 
system, b: ZW ? controlled system, and c: Ret ? controlled system. Reproduced with 
permission from [39]................................................................................................... 11 
Figure 1-3 Scheme for impedance biosensors when using (a) a modified working 
electrode or (b) two interdigitated electrodes in plane. Reproduced with permission 
from [28] ..................................................................................................................... 13 
Figure 1-4 Illustration of the pancreas with insets representative of its exocrine and 
endocrine cell types. (Image credits: Wikimedia commons). ..................................... 13 
Figure 1-5 Reduction of pancreatic trypsin and amylase output with various 
pathologies. Reproduced and modified with permission from [1], [2]. ...................... 17 
Figure 1-6 a) Trypsin and b) lipase cleaving mechanisms on peptides and 
triglycerides, respectively. .......................................................................................... 18 
Figure 1-7 (a) Process for fabricating peptide-crosslinked hydrogels over gold IDE 
sensors, specifically for sensing MMP-9 with EIS, and (b) 10 Hz impedance response 
of resulting sensor to different MMP-9 concentrations. Reproduced with permission 
from [95]. .................................................................................................................... 21 
Figure 1-8 Examples of integrated capsule systems. From left to right: PillCam 
COLON 2, GI gas sensor, and GI bleeding monitor. Reproduced with permission 
from [97]?[99]. ........................................................................................................... 23 
Figure 2-1 Schematic of collagen denaturation into gelatin. Reproduced with 
permission from [127]................................................................................................. 30 
Figure 2-2 (a) Schematic of sensor fabrication. Step 1 entailed photolithography and 
E-beam deposition of Cr/Au for 20nm/200nm, then liftoff with acetone. After 
overlaying a 75 ?m thick PVC film, 5% gelatin in buffer was spin-coated at 400 rpm 
for 30 sec, cooled overnight, then crosslinked with 0.5% glutaraldehyde solution to 
improve thermal stability. After PVC removal, gelatin was measured using a contact 
profilometer at thicknesses of 300-700 nm. (b) Microscopic image of the resulting 
sensors. ........................................................................................................................ 34 
Figure 2-3 (a) Fluidic reaction chamber for testing sensors with enzyme solutions. (b) 
Schematic of the film-coated sensor and experimental chamber (left). Photograph of 
Experimental setup (right). ......................................................................................... 36 
Figure 2-4 Resulting interdigitated finger electrode (IDE) sensors, with (Left) and 
without (Right) gelatin films, (a) pre-exposure to solutions, and post-exposure to (b) 
buffer; (c) trypsin; or (d) all three enzymes in buffer. Scale bar = 1 mm. .................. 39 
Figure 2-5 Representative SEM images of drop-cast, crosslinked gelatin films. Edge 
views of film (a) without exposure to enzyme solutions and (b) after exposure to 
trypsin, and top views of film (c) after exposure to trypsin and (d) after exposure to 
all three enzymes in buffer. ......................................................................................... 40 
Figure 2-6 Representative impedance responses at 10 kHz of sensors, either uncoated 
or coated with a gelatin film, to trypsin at 1 (both film and no film), 0.5 (film only) 
and 0 mg/mL in phosphate-buffered saline (Control). ?Trypsin?No Film? represents 
ix 
 
 
 
the impedance of PBS over the sensor, and ?Control?Film? represents the 
impedance of the film after equilibrating with the PBS. The error bars plot the 
temporal change in impedance at respective time points (temporal span = 3, n=3). .. 42 
Figure 2-7 Representative Bode and Nyquist plots for progressive exposure of (a-c) 
uncoated and (b-d) gelatin-coated sensors to 1 mg/mL trypsin in buffer. Continuous (-
) and dotted (--) lines indicate |Z| and the phase, respectively. (n=3). ........................ 43 
Figure 2-8 Representative impedance responses at 10 kHz of sensors, either uncoated 
or coated with a gelatin film, to trypsin, amylase, or lipase, each at 1 mg/mL, or 
combinations of each enzyme with trypsin. The error bars plot the temporal change in 
impedance at respective time points (temporal span = 3, n=3). ................................. 45 
Figure 2-9 Representative Nyquist plots for gelatin-coated sensors to (a) 1 mg/mL 
lipase; (b) 1 mg/mL amylase; and (c) trypsin, amylase, and lipase, each at 1 mg/mL 
over time. (n=3). ......................................................................................................... 47 
Figure 2-10 Representative QCM response of gelatin film to (a) trypsin at 1, 0.5 and 0 
mg/mL in PBS, (b) trypsin, amylase, or lipase, each at 1 mg/mL, or (c) combinations 
of each enzyme with trypsin (each at 1 mg/mL). (n=3). ............................................. 49 
Figure 2-11 Gelatin film responses to varying trypsin concentrations through analysis 
of resulting %change in in (a) mass and (b) thickness, with each sample removed 
from incubation at different time points. Error bars depict standard deviation (n=3). 51 
Figure 2-12 Logarithmic rate of reactions of gelatin films in solutions with varying 
pancreatic trypsin concentrations, calculated using the slopes in percent change of 
film mass and thickness. Error bars depict standard deviation (n=3). ........................ 52 
Figure 2-13 (a) Equivalent circuit model and (b) schematic for thin films over 
impedance sensors. Reproduced with permission from [39]. ..................................... 54 
Figure 2-14 (a) Resulting gelatin film thicknesses and different spin speeds, both pre- 
and post- rinsing with DIH2O. (b) Top-level view of example sensor outline, with 
each sensor label. (c) Resulting film thicknesses for each spin speed, broken down by 
sensor ? using labels from (b) ? and rinsing. Error bars depict standard deviation 
(n=3). ........................................................................................................................... 56 
Figure 2-15 (a) Series equivalent circuit mode of LCR meter. (b-c) Electrical 
(capacitive and resistive, respectively) analysis of sensors across varying electrode 
spacings and comparing between the presence and absence of a film. Error bars depict 
standard deviation (n=3). ............................................................................................ 58 
Figure 2-16 Regression analysis of different gelatin film thicknesses and their impacts 
on sensor (a) capacitance and (b) resistance across 25, 50, and 75 ?m electrode 
spacings. ...................................................................................................................... 60 
Figure 3-1 Modules contained within endoscopy capsule: (1) Optical Dome, (2) LED, 
(3) Short-focus lens, (4) CMOS image sensor, (5) RF model, (6) MCU, and (7) Power 
model [176]. ................................................................................................................ 66 
Figure 3-2 Human GI tract morphology with respective pH ranges. Reproduced with 
permission from [184]................................................................................................. 68 
Figure 3-3 Depiction of system outlook and application (Image credits: National 
Cancer Institute). Polymer coatings are intended to dissolve at either the stomach (pH 
1.5-3) or the duodenum (pH 7-8.5). At the duodenum, gastric acid-neutralizing 
bicarbonate is secreted from the pancreas via the sphincter of Oddi. ......................... 70 
x 
 
 
 
Figure 3-4 System electronics overview. Circuit design includes a multiplexer (MUX) 
integrated circuit (IC), a capacitance-voltage-converter (CVC) IC, a 1.5 MHz step-up 
DC-DC converter, and a BLE system-in-package (SiP) micro-controller unit (MCU).
..................................................................................................................................... 72 
Figure 3-5 (a) Circuit calibration with known capacitors across four multiplexed 
inputs. The resulting system senses capacitive changes in the 0.8-220 pF dynamic 
range with a sensitivity of 3.2x10-3 pF/mV and operated using a 3.3 V source. (b) 
Data comparison between Bluetooth versus serial (UART) communication in 
different experiments, where the capsule (open gratings) was measured in air, inserted 
into buffer, then removed. Error bars depict standard deviation (n=4). ..................... 72 
Figure 3-6 (a) CAD model of the capsule used with external circuit. (b) 3-D printed 
version with zoomed up view of gratings. Scale bar = 1 mm. .................................... 73 
Figure 3-7 Cross-sectional optical images of the sampling gratings (a) uncoated and 
(b) with 1 coating of polymer, where yellow arrows indicate the grating locations and 
yellow boxes indicate the inlets. (c) Characterization of polymer coating thickness vs. 
number of dipping steps from both non-grating and grating regions over the 3-D-
printed capsule surface. Error bars depict standard deviation (n=3). Scale bar = 1 mm.
..................................................................................................................................... 74 
Figure 3-8 (a) PCB rendering with sensor die fixated between designated pads using 
epoxy. Contact was made by curing conductive silver ink. (b) Schematic of the 
assembled capsule setup. (c) Calibration of BLE MCU at the PCB level using 
standard known capacitors. (d) Photograph of assembled capsule. Scale bar = 5 mm 
unless otherwise specified........................................................................................... 76 
Figure 3-9 Sensor responses (inserted into buffer at 5 min) of capsules with different 
numbers of S 100 coatings (0, 2, 3, and 4, respectively) with increasing pH using 
titers of 1 M NaOH. Error bars depict standard deviation (n=4). Chamber filling is 
characterized by a capacitance change of ~50 pF from the initial capacitance in air. 
For each formulation, it was determined that 3 coatings were optimal for the w/v% 
used. ............................................................................................................................ 79 
Figure 3-10 Sensor responses inserted into capsules with different types of coatings (3 
dip-coatings each) compared to uncoated controls: (a) L 100, (b) E PO, and (c) 
combined coatings, where E PO is outer-most and L 100 is be-tween the 3-D-printed 
capsule and the E PO layers. Error bars depict standard deviation (n=4). (d) Capsule 
rendering for displaying coating conditions for each sequence. ................................. 82 
Figure 3-11 Real-time representative trials of capacitance measurements obtained via 
BLE from the 3x coated capsule immersed in control (neutral, pH 7) (red) and 
specific target (blue) pH (acidic, pH 3). The rapid increase in capacitance represents 
buffer inflow and contact with the sensors, due to coating dissolution. (n=2). .......... 84 
Figure 3-12 Progression of polymer coating throughout (a) control and (b) pH-based 
dissolution tests. Beginning, 30 minutes, and 1 hour immersion periods are from the 
left, middle, and right, respectively. Scale bar = 3 mm. ............................................. 84 
Figure 3-13 (a) Diagram illustrating the regions along each axis where pressure was 
applied along the load frame. The colors are used for reference in (b) the applied 
force profiles for each setting. The failure point for each is indicated with each black 
circle. (c-d) Representative photographs of the capsule package upon reaching failure 
more for ?vertical? and ?on side? setups, respectively. (n=1 for each condition). ..... 86 
xi 
 
 
 
Figure 4-1 (a) Lipase-induced hydrolytic digestion of stearin into glycerol and three 
stearic acid molecules. (b) Bile salt-induced emulsification of triglyceride globules 
into micelles. ............................................................................................................... 94 
Figure 4-2 Deposition strategies for obtaining TG films over sensors: (a) drop-casting 
and (b) dip-coating of molten TG solution while the substrates were either left at 
ambient or pre-heated to TG melting temperature for 5 minutes. (c) Dip-coating of 
assembled capsules into dyed pH-sensitive copolymer solutions to form GI-targeting 
coatings, repeated for sequential coatings. ................................................................. 95 
Figure 4-3 Assembly process for platform. Steps 1-4 used for sensor characterization 
of TG-SPAs with connection of VDD and GND to external power supply, while 
Steps 5-8 are used for testing complete capsule. At Step 5, after connection to 
Lithium polymer battery. Deep Sleep mode is entered for a 3-hour time period via 
transmitted signal to allow for curing of adhesion and coating materials at respective 
steps. After Step 8, capsule electronics enter Active mode to enable capacitance 
measurements and signal transmission to mobile phone. ........................................... 96 
Figure 4-4 (a) Schematic depicting normal GI pH progression along with adjacent 
organs. (b) Experimental setup for complete capsule characterization. ..................... 98 
Figure 4-5 Resulting thicknesses in films with each deposition strategy and initial 
SPA temperature. Error bars depict standard deviation (n=3). ................................. 102 
Figure 4-6 Representative sensor response testing after insertion of TG-SPAs into 
buffer solution (0.1 M PBS) when TG film composed of either pure stearin or 
stearin:glycerol (S:G) ratios of 2:1, 1:1, or 1:2. Film most stable, i.e. least response 
over time, with the 2:1 S:G ratio compared to other compositions. (n=2). .............. 103 
Figure 4-7 Sensing in the presence of different biochemical species at varying 
concentrations. (a) PL, (b) BAs, and (c) pancreatic trypsin. ..................................... 105 
Figure 4-8 Representative SEM images for characterization of SG films after 
exposure to various solutions at 30- and 60-minute time intervals. (n=3). .............. 108 
Figure 4-9 (a) Combined pH sampling and enzyme sensing in integrated capsule. 
Measurements were initiated after immersion into acidic solution for 25 minutes 
(blue), after which the capsules were removed and immersed into a neutral solution 
(red), where a spike was observed around 19 minutes, indicated by the circle. Finally, 
the capsule was removed and immersed into a solution containing 1 mM lipase 
(black), where a distinct increase in capacitance is observed, indicated by the arrow. 
(b) Photographs taken of the capsule at the end of each respective sequence, 
designated by the label color: b-i: acid condition, b-ii: neutral condition, b-iii: lipase 
condition, and b-iv lipase condition after removal from the solution. (n=1). ........... 110 
Figure 5-1 Depiction of application. The capsule is protection in environmental pH 
reflecting that of the stomach until reaching the small intestine via pH-sensitive 
polymer coating. Upon dissolution, the fluid is sampled via embedded gratings, and 
pancreatic lipase reacts with triglyceride coatings to increase the capacitance of the 
sensors. This is transmitted to a phone wirelessly, and the data can be shared with 
medical practitioners. ................................................................................................ 117 
Figure 5-2 (Left) Schematic and (right) optical images of sensors with deposited film. 
The optical images indicate distinct removal of film after trypsin response compared 
to buffer alone, where the film remains relatively intact. ......................................... 118 
xii 
 
 
 
Figure 5-3 (a) Top and (b) bottom views of sensor-connected PCB, and (c) 
photograph of assembled capsule. Scale bar=5 mm. ................................................ 119 
Figure 5-4 Peristalsis simulator, consisting mostly of 3D printed components. ...... 124 
Figure 5-5 SEM imaging of (a) porcine (reproduced with permission from [246]) and 
(b) 3D-printed villi with an EnvisionTEC Perfactory 4. (c) Implantation capsule robot 
(reproduced with permission from [249]). (d) 3D-printed anchoring concept for 
passive capsule implantation; left inset: head of Taenia Solium Scolex (reproduced 
with permission from [248]), right inset: close-up of microhook mimic. ................ 126 
Figure 5-6 Physiology of GI affected regions, reflecting potentially relevant 
information accessible by capsule devices. .............................................................. 127 
Figure 6-1 IDE sensors used in chapter 2. Finger width: 2 ?m, finger spacing: 4 ?m, 
finger length: 1 mm, number of fingers per electrode: 65. ....................................... 130 
Figure 6-2 IDE sensor patterns used in chapter 3 to measure integrity of the 
packaging coatings. (a) Complete photomask with 10 die for each 4x1 sensor die, 
each with different finger geometries: 25, 50, and 75 ?m width and spacing and 44, 
22, and 14 fingers per electrode, respectively. Finger length: 2 mm. ....................... 131 
Figure 6-3 IDE sensor patterns used in chapters 3 and 4 for SPAs. Mask Design: 124 
Sensor Output, 62 each of long (length=3 mm) /short (1.5 mm) versions. Number of 
fingers per electrodes = 40. ....................................................................................... 132 
Figure 6-4 (a) Cross-section of simulation model (top to bottom: medium, film, 
sensor, wafer), (b) mean capacitance from experimental degradation of film over IDE 
sensor (left y-axis) and simulation (right y-axis) results. Between electrode fingers, 
electric field intensity increases significantly in the presence of gelatin. Electric field 
height decreases significantly, the magnitude remains the same, and lateral electric 
field distribution is altered. ....................................................................................... 135 
Figure 6-5 MATLAB GUI. Up to four data sequences can be plotted, three of which 
can be plotted individually on the right while all are overlaid in the plot on the bottom 
left. Features that can be modified directly in the GUI are smoothing filters, axes 
limits, and legend labels. ........................................................................................... 151 
Figure 6-6 Depiction of data format in .csv file. This format is currently programmed 
into the app, while the algorithms to process the capacitance data are embedded in the 
MATLAB code above............................................................................................... 151 
 
xiii 
 
 
 
List of Abbreviations 
AC: Alternating current 
ADC: Analogy-digital converter  
BA: Bile acids 
BAEE: N?-Benzoyl-L-arginine ethyl ester  
BLE: Bluetooth Low Energy  
CA: Carbohydrate antigen  
CCK: Cholecystokinin 
CEA: Carcinoembryonic antigen  
COTS: Commercial off the shelf 
CT: Computed tomography  
CTC: Circulating tumor cells  
CVC: Capacitive-voltage converter 
ctDNA: Circulating tumor deoxyribonucleic acid  
DC: Direct current 
DI: Deionized 
DNA: Deoxyribonucleic acid 
EGD: Esophagogastroduodenoscopy 
EIS: Electrochemical impedance spectroscopy  
ENPD: Endoscopic nasopancreatic drainage 
EPI: Exocrine pancreatic insufficiency  
ERCP: Endoscopic retrograde cholangiopancreatography  
ERP: Endoscopic retrograde pancreatography  
FC: Film-coated chip 
xiv 
 
 
 
FDA: Food and Drug Administration 
FDM: Fused Deposition Modeling 
FET: Field-Effect Transistor 
FFA: Free fatty acids 
FNA: Fine-needle aspiration 
GI: Gastrointestinal  
GL: Gastric lipase 
GUI: Graphical user interface 
IBD: Inflammatory bowel disease  
IBS: Irritable bowel syndrome  
IC: Integrated circuit  
IDE: Interdigitated electrodes  
MCU: Microcontroller unit  
MEMS: Microelectromechanical systems 
MMP: Matrix metalloproteinase 
MRCP: Magnetic resonance cholangiopancreatography  
MUX: Multiplexer 
PA: Pancreatic adenocarcinoma 
PBS: Phosphate buffered saline  
PCB: Printed circuit board 
PET: Positron emission tomography  
PFT: Pancreas function testing 
PL: Pancreatic lipase 
xv 
 
 
 
PVC: Polyvinyl chloride 
QCM: Quartz crystal microbalance  
RF: Radiofrequency 
RNA: Ribonucleic acid  
RPM: Revolutions per minute  
SAW: Surface acoustic wave  
SEM: Scanning electron microscopy 
SiP: System in package  
SG-SPA: Stearin and glycerol coated sensor-printed circuit board assembly 
SPA: Sensor-printed circuit board assembly 
TAME: p-toluene-sulfonyl-L-arginine methyl ester 
WSTK: Wireless starter kit
xvi 
 
 
 
1. Chapter 1: Introduction 
1.1 Background and Motivation 
The gastrointestinal (GI) tract is a complex canal that, while composed of 
distinct organs with respective functions, hosts a wide range of biochemical and 
biomechanical interactions critical to homeostasis. To maintain various organ 
functions, there is a continuously evolving dependence on understanding the functional 
status at any given time through GI diagnostics in particular, which range broadly 
depending on the suspected condition, how it is known to manifest in physiological 
processes, and its severity.  
One example of a physiological process that can require monitoring ? and is a 
major focus of this dissertation ? is exocrine pancreatic function, which is critical to 
maintaining one?s nutritional intake at the very least, but is often compromised in a 
variety of pancreatic pathologies, including pancreatitis, pancreatic adenocarcinoma 
(PA), and cystic fibrosis [1]?[3]. While various aspects of human physiology and the 
procedures involved are well understood by physicians and medical researchers, 
current diagnostic tools reliant on biomarkers ? objective medical indicators of a 
physiological state ? suffer from a combination of limited specificity or sensitivity, 
high cost and invasiveness, or only detect the condition severity at the point at which 
efficacy of traditional therapies has diminished [4], [5]. Though biomarkers vary with 
condition, enzymes, which are a class of proteins naturally synthesized in cells to 
catalyze (or accelerate) biochemical reactions, have been demonstrated as 
physiological indicators and remain a major category of biomarkers [6]. There are 
further subclasses of enzymes required for food digestion and nutrient absorption in the 
1 
 
 
 
GI tract: proteases (trypsin, chymotrypsin, and elastase), lipases (steapsin), and 
amylases (?-amylase). These enzymes are secreted in either active forms or precursors 
(zymogens) by the pancreas in a bicarbonate solution, or buffer, that enters the 
gastrointestinal (GI) tract ? into the duodenum ? by way of the pancreatic duct [7]. 
Additional markers such as carcinoembryonic antigen (CEA), carbohydrate antigen 
(CA) 19-9, or specific microRNAs have been found present in these secretions that 
correlate with early stages of PA [8], [9]. 
Recently, there has been a surge in advanced microsystem development driving 
innovation for ingestible sensors capable of replacing or aiding invasive clinical 
methods [10]?[12]. Such devices are designed to reveal new information about 
physiological responses to various stimuli (discussed in detail in Section 1.3.4), 
depending on the sensor types and the materials used to modify them [13]. However, 
none have been used to measure enzyme activity, which can be invaluable to expanding 
their utility. Moreover, consistent with Moore?s law, microelectronics have advanced 
to point where circuit footprints can cost-effectively be fabricated in reduced sizes 
while increasing complexity [14]. The primary challenge remains how to leverage this 
for interfacing with biomaterials such that we can measure physiological processes in 
regions where smaller systems are at a great advantage for obtaining previously 
inaccessible data. This challenge is augmented by the chaotic nature of the GI tract, 
especially for integrating the microelectronics into the ingestible, autonomous devices 
that must navigate through it [15]. 
Distribution of specific tasks, such as safe navigation and specific sensing, to 
different modules is critical for embedding functionality in complex systems and 
2 
 
 
 
devices [16]. Modular integration can be difficult when each requires a different 
fabrication method, hence the growing utility of more custom tools such as the variety 
that fall under the category of 3D-printing, including stereolithography, photopolymer 
jetting, laser sintering, and fused deposition modeling (FDM), among others [17]. 
Furthermore, features such as biocompatibility and biodegradability have become 
common and almost essential themes in many bioelectronics devices for providing safe 
interfaces with other biologics and embedding sustainability, respectively [18]?[21]. 
This doctoral research aims to overcome these challenges and demonstrate the 
development of a wireless integrated capsule microsystem for navigating the GI tract, 
sampling pancreatic secretions, and monitoring enzyme activity in situ for indicating 
pathological conditions. I anticipate that this system will lay the foundation for hybrid 
fabrication and integration methods toward measuring a broader range of GI targets. 
1.2 Summary of Accomplishments 
In this work I leveraged biochemical properties of various materials as coatings 
individually designated for (1) packaging our system to sample fluid with a specific pH 
characteristic of a target GI region and (2) insulating our sensors until degrading in 
response to catalytic pancreatic enzymes. Specifically, I integrated these features into 
an ingestible capsule designed for ingestion and autonomously navigating through the 
GI tract via natural peristaltic forces for sampling pancreatic secretions, measuring 
pancreatic enzyme activity, and wirelessly transmitting the data via smartphone 
interface for external analysis. I envision using this platform as a system for not only 
targeting specific regions in the GI tract by utilizing a variety of coating materials, but 
for sensing different enzymes that might be in those regions. This broadens the 
3 
 
 
 
applicable conditions for which this device could be used, limited only by current 
understandings of how GI secretions vary with these conditions. Such a platform would 
minimize efforts and intermediaries between the physician and the source of a patient?s 
pathology, with the potential for aiding in differential diagnostics. Toward enabling 
this research, three primary objectives were investigated: biomaterial-based electrical 
transduction systems for measuring duodenal enzyme activity, packaging materials for 
system protection and GI region targeting, and integrating these systems to function 
simultaneously into an independent device. 
1.2.1 Investigation of Biomaterial Film-Based Electrical Transduction Systems for 
Measuring Enzymatic Activity 
This work begins with the investigation of thin films made from biomaterial 
substrates that respond preferentially to health-related enzymes of interest. In this part 
of the thesis, I developed a variety of deposition parameters for obtaining thin films of 
gelatin and their subsequent crosslinking over both wafer- and individual sensor-level 
substrates. The protocols developed here are the first time of not only depositing 
glutaraldehyde-crosslinked gelatin films over impedance sensors and the 
characterization of the impact of different spin-coating parameters on their resulting 
thicknesses, but determining and confirming the change in various impedance 
elements, including resistance and capacitance, with respect to these varying 
thicknesses. Furthermore, we established relationships between the rate at which these 
film thicknesses, and ultimately impedance, change in response to local concentrations 
of a pancreatic trypsin (1 ?g/mL ? 1 mg/mL). The impact of these methods and 
characterizations is that they enable further work on integrating similar films, i.e. those 
4 
 
 
 
from either gelatin-based hydrogels or other biomaterials, with traditional MEMS 
technology and utilizing them for more advanced developments in enzyme detection 
through impedance sensing.  
1.2.2 Investigation of Embedded Packaging Materials for Gastrointestinal Targeting 
Due to naturally occurring pH gradients in GI regions and their contents, 
polymer coatings commonly used in the pharmaceutical industry for dissolving in a 
specific pH, provides a mechanism for GI targeting [22]?[24]. This aim resulted in the 
development of a microelectronics sensing platform that leverages these polymers as 
coatings in combination with a biocompatible 3D-printed capsule structure as a hybrid 
package to protect internal system elements from nonspecific environmental stimuli. 
The platform was designed using off-the-shelf components to acquire changes in sensor 
capacitance for downstream utility in measuring changes in biomaterial film thickness 
from enzyme reactions as described in Aim 1. This capacitance measurement is then 
transmitted wirelessly using Bluetooth Low Energy (BLE) for real-time measurement 
control, realized through the use of a facile user interface via an Android app. Using 
similar methods used for coating therapeutic capsules in drug delivery applications, I 
adapted a dip-coating protocol for layering the pH-sensitivity polymers over our 3D-
printed capsules, which allow a sensor-sample interface using microfluidic gratings 
that allow fluid inflow, while subsequently developing an assembly process that 
maintains functionality of the individual modules (i.e. sensors, power supply, 
electronics, etc.) required for autonomous operation of the device. These processes 
serve to promote the targeting capabilities of future ingestible devices while offering 
insight as to interfacing MEMS sensor signals with users such as patients or clinicians.  
5 
 
 
 
1.2.3 Development of Ingestible Integrated System for Targeted Gastrointestinal 
Sampling and Enzyme Sensing 
 The third accomplishment of this work is the development of triglyceride film-
based deposition and sensing strategy for detecting biochemical species present in the 
duodenum and its subsequent integration into a GI-targeting capsule system that can 
sample simulated fluid in the duodenum. We leverage deposition methods such as dip-
coating and drop-casting for triglyceride films over the sensor surfaces utilized in Aim 
2, and determine their respective limitations in comparing process parameters such as 
film composition and substrate preparation. This is the first demonstration of utilizing 
a purely capacitance-sensing platform for measuring reactions between triglyceride 
films, specifically from stearin and glycerol, and duodenal analytes such as pancreatic 
lipase or bile acids. Furthermore, we are the first to reveal morphological changes of 
such films after undergoing hydrolytic and emulsification reactions to lipase and bile 
acids, respectively, compared to buffering fluid alone or with local trypsin as a 
nonspecific analyte. The assembly process used for integrating the capsule modules in 
Aim 2 is modified further to maintain reactive films over the sensor surfaces for 
subsequent testing in the pH-soluble polymer-coated capsules, establishing various 
required considerations and validating the feasibility of pH-targeted sampling and 
sensing. The successful development of this device will enable further advances in both 
novel hybrid fabrication strategies for integrated device development and exploratory 
endeavors for GI evaluation with ingestible microsystems. 
6 
 
 
 
1.3 Literature Review 
 This section provides the necessary background and literature review as 
applicable to the presented work. First, I will discuss previous work on electrical 
sensing systems for biological species and their requirements for successful operation.  
Next, I will present a review on the current state of traditional GI diagnostic tools with 
an emphasis on pancreatic health. I will then expand on pancreatic enzyme and 
substrate interactions while introducing similar species present through the GI tract. 
Finally, recent advances in ingestible systems for addressing challenges in GI health 
will be evaluated, introducing the rationale for various design choices made during this 
work. 
1.3.1 Electrical Biosensors 
 In general, biosensors are devices that convert the presence of biological 
molecules or activity into an observable signal [25]?[27]. Biosensors are often 
designated to measuring biomarkers, which, in a clinical setting, are indicators for 
normal or pathological function of physiological processes. Figure 1-1 illustrates the 
general mechanism for a biosensor. The most common signals processed in biosensors 
are electrical, optical, mechanical, and magnetic, though this dissertation will 
emphasize efforts for interfacing biomolecules on electrical biosensors. Their 
performance can be defined through the following criteria: sensitivity, working and 
linear concentration ranges, detection limits, selectivity, reliability, response time 
(steady-state or transient), sample throughput, reproducibility, stability, and lifetime 
[28]. 
 
7 
 
 
 
 
Figure 1-1 Variety of biosensing mechanisms [29]. 
Electrical or electrochemical biosensors rely on a biological system?s ability to 
directly alter an electrical signal, generally in the form of a simple applied voltage or 
current [30], [31]. These sensors are popular as they do not require intermediate 
transduction mechanisms, can be easily miniaturized for subsequent integration with 
most microelectronic systems, are low cost to produce and, depending on the structure 
can be fashioned to be mechanically robust. Additionally, while innovations in sensing 
generally aim to expand the detection limits and ranges, the concentrations of the 
targets of interest, discussed more in section 1.3.2, are relatively higher, allowing the 
use of simpler systems. The signals are frequently measured as either impedimetric or 
conductometric (interdigitated, metal electrodes), potentiometric (ion-selective, glass, 
gas, or metal electrodes), amperometric (metal, carbon, or chemically modified 
electrodes), or field-effect transistor (FET, ion or enzyme sensitive), and induce an 
increase or decrease in signal with respect to a change in surface characteristic from a 
bio-functionalized sensor material [28]. The medium of applied excitation is desired to 
8 
 
 
 
possess a certain threshold of conductivity that would enable electron transport from 
the source to receiver, hence the use of metals such as gold or platinum as the path of 
travel.  
One primary challenge of these sensors is determining the appropriate method 
for interfacing the material of the transducer ? typically an electrode ? with the 
biological materials. The latter can include proteins, carbohydrates, lipids, nucleic 
acids, or even just ions released from a physiological function. One of the most 
ubiquitous clinical biosensors is the glucose sensor, which operates amperometrically 
by relating change in current to the amount of enzyme catalysis between glucose and 
glucose oxidase, the latter of which is the material bound to the sensor surface. The 
reaction, based on reduction-oxidation, results in electron generation, thereby 
increasing the current at the peak potential of an electrochemical cell. The peak 
potential is specific to the experimental condition, though the current can be 
approximated by the Nernst equation (1): 
?? (Ox)
? = ?0 + ??  (1) 
?? (Red)
 
Here, E is the cell potential, E0 the standard potential of a species, R the universal gas 
constant (8.314 J/(mol?K)), n the number of electrons, T the temperature, and (Ox) and 
(Red) the relative activities of the oxidized and reduced analytes at equilibrium [32]. 
This is used to determine concentrations of electrochemical activity, and is popular 
among other electrical biosensing strategies due to its high sensitivity, compatibility 
with microfabrication processes, minimal power requirements, and economical cost, 
among others [25]. These systems require two or three electrodes. Both require (1) a 
9 
 
 
 
working electrode for responding to the target analyte and (2) a counter electrode to 
balance the reaction. Three-electrode systems possess an additional reference electrode 
to maintain a constant potential independent of the solution properties.  
 Direct electron transfer is the ideal scenario for electrochemical sensing systems 
as it results in simpler designs. However, this only occurs with species that already 
possess a charge, including ions. Some sensors require a secondary substrate; using the 
glucose sensor as an example, glucose oxidase is required to generate the electron via 
catalytic reaction. In other scenarios, mediators are implemented that react with the 
electrode, such as ferrocyanide or ferrocene derivatives, that avoid interfering with 
other reactants in the solution and possess lower potential requirements [33]. Both 
options lead to new design requirements in the system, i.e. methods to ensure the 
secondary substrate or mediator is constantly within range or access of the electrode 
surface. While this is not a problem with most benchtop setups, integrated systems that 
rely on isolated sensing chambers must adapt methods for ensuring their containment. 
One purpose of mediators is their ability to amplify reactions as a means of embedding 
specificity to the sensor. Another method toward this end is the utilization of 
membranes that filter interfering species access to the sensor surface or provide an 
additional layer for substrate immobilization. For example, nitrocellulose filters are 
fabricated with pores that can reach 0.22 ?m, which is permeable to fluid flow to 
underlying sensors while maintaining protein-retaining properties [34]. Further, lipid 
bilayer membranes have been utilized for mimicking physiological conditions [35]. In 
addition to adding further fabrication and design steps, however, such membranes are 
10 
 
 
 
often at risk of fouling or pore saturation, thereby requiring solutions for mitigation 
such as treatments that increase hydrophobicity [36].  
In contexts where no redox species are involved and changes in electrical 
signals at different frequencies, i.e. AC systems, adds value to understanding a 
mechanism of reaction, electrochemical impedance spectroscopy (EIS) is another 
viable sensing technique. EIS, which can be categorized into Faradaic and non-Faradaic 
subgroups based on their respective requirement and absence of a reference electrode, 
offers information on changes in the chemical interface of modified electrodes, 
modeled specifically using the Randles and Ershler equivalent circuit, as well as 
reaction rates through sequential spectra [37], [38]. The model for Faradaic EIS is 
displayed in Figure 1-2, while the resulting equation is as follows: 
 
 
 
Figure 1-2 Randles-Ershler equivalent circuit models for (a) Faradaic and (b) non-Faradaic EIS. (c) 
Nyquist spectra for Faradaic EIS for a: ZW ? and Ret ? controlled system, b: ZW ? controlled system, and c: 
Ret ? controlled system. Reproduced with permission from [39]. 
??? ???
2
???? (2) 
?(?) = ?? + ? = ?
? + ???? 
1 + ?2?2 ?2?? ? 1 + ?
2?2 ?2?? ?
Here, Z is the total impedance, Rs and Rct (shown in Figure 1-2 as Ret) are the solution 
and charge transfer resistance, respectively, Cd is the double-layer capacitance and ? is 
11 
 
 
 
the frequency. ZW in the figure represents the Warburg impedance, which is the result 
of ion diffusion from the solution to the electrode surface. The equation identifies the 
total impedance as two distinct parts: Z? and Z??, which are the real and imaginary 
impedance, respectively. These are often analyzed by plotting against one another to 
form Nyquist plots, which produce spectra that offer information on the individual 
circuit elements [39], [40]. This can provide insights as to biochemical events occurring 
at the electrode surface, depending on which circuit element is affected. This topic will 
be further discussed in relation to this work in chapter 2. 
 As discussed earlier in this section, electrical and electrochemical-based 
measurement systems are well suited for integration with microelectronics systems, 
which makes them applicable for point-of-care and, as demonstrated throughout this 
thesis, capsule-based systems. There are many that are already used for measuring a 
variety of biomarkers, including pathogens and toxins, in the clinical setting, though a 
major limitation of EIS is generally the lack of labelling processes involved, which 
reduce sensitivity and performance in complex media such as blood. However, they 
continue to increase in popularity and efforts remain to improve surface and binding 
strategies of targets. A general schematic of how the sensors are arranged is depicted 
below, while more information on specific impedance-based biosensors and their 
respective performances can be reviewed in other works [41], [42]. 
12 
 
 
 
 
Figure 1-3 Scheme for impedance biosensors when using (a) a modified working electrode or (b) two 
interdigitated electrodes in plane. Reproduced with permission from [28] 
1.3.2 Exocrine Pancreatic Health and Diagnostics 
 
Figure 1-4 Illustration of the pancreas with insets representative of its exocrine and endocrine cell types. 
(Image credits: Wikimedia commons). 
This section will describe the clinical topic of interest in designing our 
biosensors. The human pancreas, illustrated in Figure 1-4, is a focus in this dissertation 
due to its relationship to the GI tract and the general difficulty associated with its 
various diagnostics. It is a gland comprised of an endocrine and exocrine system, the 
13 
 
 
 
latter of which is primarily responsible for secretion of digestive enzymes and sodium 
bicarbonate into the duodenum, via the sphincter of Oddi, for macromolecular 
fragmentation and gastric acid neutralization, respectively. The most common 
pathologies that affect these secretions are pancreatitis (acute and chronic) and various 
forms of pancreatic cancer. A lack of digestive enzymes, however, is more generally 
referred to as exocrine pancreatic insufficiency (EPI), and is often present in cystic 
fibrosis, a primarily endocrine affected pathology [3], [43]. Before diagnosis of these 
conditions however, a patient must present various symptoms as part of the screening 
process as part of the differential for identifying the exact condition. For example, both 
pancreatitis and pancreatic cancer can cause upper abdominal pain, jaundice, fever, 
and/or unexplained weight loss, but further tests are required to verify both the 
underlying issue as well as the severity of the condition [44], [45]. 
Current diagnostic strategies for evaluating the pancreas can range from direct 
analysis of the tissue itself or indirect analysis via its secretion products. Direct analysis 
is generally the most specific, consisting of imaging techniques for physical 
abnormalities and blood testing or tissue biopsy for molecular and genetic testing. 
Imaging tools are gold standards for verifying the presence of a tumor or neoplasm 
(tumor precursor), or even enlargement or inflammation [4], [46], [47]. Specific 
strategies include abdominal computed tomography (CT), endoscopic ultrasound 
(EUS), magnetic resonance cholangiopancreatography (MRCP), magnetic resonance 
imaging (MRI), endoscopic retrograde cholangiopancreatography (ERCP), or positron 
emission tomography (PET) [47]?[49]. The primary challenge with these strategies for 
cancer, unfortunately, is their lack of efficacy in early-stage detection, considering the 
14 
 
 
 
images are analyzed and diagnosed by physician ability. EUS is often used to guide 
biopsy through fine-needle aspiration (EUS-FNA), thereby offering further molecular 
information but still suffering from high invasiveness. ERCP, while useful for 
collecting samples for cytological analysis, can occasionally induce pancreatitis, 
bleeding, and cholangitis [50].  Computer-aided approaches with EUS are also being 
developed to improve diagnosis rates, and while they have increased performance rates 
there remains room for improvement [51]. Pancreatitis, alternatively, requires 
significantly less effort to diagnose. Digestive enzymes such as amylase, lipase, and/or 
trypsin, for example, can be significantly elevated in serum, while the fecal elastase 
test is the gold standard when diagnosing severe chronic pancreatitis [4]. The primary 
blood marker that is a gold standard for identifying the presence of cancer is 
carbohydrate antigen 19-9 (CA 19-9), though it is not specific to pancreatic cancer and 
has limited sensitivity for smaller tumors [47], [49], [52], [53]. Other markers include 
circulating tumor DNA (ctDNA) and cells (CTC) analyzed for KRAS mutations, which 
can often be used to determine the progression of treatment as well. Unfortunately, 
these are very difficult to isolate and identify, which only becomes more challenging 
in the early stages [54]. 
Pancreatic juices have been investigated as a source of diagnostic information 
[8], [47], [49], [55]?[57]. Duodenal aspirates have been analyzed for comparing various 
pancreatic pathologies for differentiation as they have been least physiologically 
processed and therefore less likely to be contaminated or the contents degraded from 
spontaneous autolysis, especially for DNA or RNA markers [58]?[60]. The region of 
secretion in the duodenum is targeted using esophagogastroduodenoscopy (EGD), 
15 
 
 
 
endoscopic retrograde pancreatography (ERP), or endoscopic nasopancreatic drainage 
(ENPD) [57], [61]. Further, genetic and protein analysis has yielded markers such as 
K-Ras, Smad4, KL-6, PKD, telomerase activity, DNA methylation, CA 19-9, or 
microRNA?s in the juice for cancer, some of which allow for concrete differentiation 
from pancreatitis [8], [53], [57], [62]. Various analyses have shown differences in 
digestive enzyme activity from the secretions [1], [63]?[65]. Figure 1-5 illustrates 
qualitative trends in enzyme output, impacted by various pathologies, as this will be 
emphasized more throughout this work. Overall, however, sample extraction and 
analysis can be cumbersome while ultimately losing molecular integrity, thereby 
driving the need for in situ measurements for a combination of several biomarkers 
discussed above that can be achieved with tools that require less human intervention 
such as  ingestible capsule technology. With tools that can achieve measurements in 
situ, there is greater potential to detect biomarkers at an earlier pathological stage to 
prevent transit-related sample degradation, while allowing a higher throughput of non-
invasive and likely less costly-procedures with less involvement from clinical and 
laboratory personnel. Furthermore, in situ sensing offers the ability for real-time 
analysis upon data acquisition, enabling a more rapid response for informed decision 
making. This will be elaborated more in chapters 3-5. 
 
 
 
 
 
16 
 
 
 
 
 
Figure 1-5 Reduction of pancreatic trypsin and amylase output with various pathologies. Reproduced 
and modified with permission from [1], [2]. 
1.3.3 Pancreatic Enzyme Activity and Other Gastrointestinal Fluid Contents 
Detecting the presence of multiple molecular markers in pancreatic secretions 
has the potential to improve the differential diagnosis for pancreatitis or screening for 
pancreatic adenocarcinoma. Depending on the sensing strategy utilized, accuracy and 
precision can be optimized to compete with the previously mentioned strategies, 
limited only by recent advances in technology.  
17 
 
 
 
(a) (b)  
Figure 1-6 a) Trypsin and b) lipase cleaving mechanisms on peptides and triglycerides, respectively. 
Many pancreatic enzyme activity biosensors are under investigation that use 
substrates responsible for hydrolysis with the target enzyme. Trypsin, for example, 
cleaves peptide chains on the carboxyl ends of amino acid residues lysine and arginine 
unless proline is present [66], [67]. The mechanism is illustrated in Figure 1-6. 
Substrates used in standardized or reported assays are N?-Benzoyl-L-arginine ethyl 
ester (BAEE) [68] and p-toluene-sulfonyl-L-arginine methyl ester (TAME) [69]. 
Alternatively, sensors have used poly-l-lysine, cytochrome c, or gelatin [70]?[73]. 
Pancreatic lipase, however, cleaves triglycerides at ester bonds to form glycerol and 
three fatty acids. This occurs specifically when the enzyme undergoes interfacial 
activation, which consists of the enzyme active site becoming uncovered at a lipid-
water interface [74]. Substrates used in assays or sensors include olive oil, composed 
4-30% of triolein which can also be used as a substrate, while an amperometric sensor 
has been reported that uses glycerol dehydro-genase/NADH oxidase. Pancreatic ?-
amylase cleaves at alpha-bonds of large polysaccharide chains to produce mono- and 
disaccharides glucose and maltose, respectively [7]. Therefore, glycogen or starch in 
18 
 
 
 
various forms such as amylose or amylopectin are used as substrates in assays or in 
biosensors for determining amylase activity, which can potentially enable the detection 
of amylase using pre-existing redox sensors using such materials [75], [76]. 
 
The resulting hydrolytic events induce structural changes that can be sensed 
with a variety of transduction methods, such as electrochemical, piezoelectric, optical, 
thermoelectrical, or photosensitive, among others [77]. Several studies use a quartz 
crystal microbalance (QCM) to sense changes in crystal resonant frequency as the 
substrate is degraded in response to varying concentrations of the enzymes mentioned 
above [72], [76], [78]. These studies consisted of depositing films of the biomaterials 
on the sensor surface, and in some instances, would be embedded with nanoparticles 
or chemical cross-links to enhance sensitivity or decrease non-specific binding, 
respectively. In other sensors, these films can act as insulators, coating surfaces and 
changing their electronic characteristics, such as conductivity or dielectric permittivity, 
depending on the material. For example, a pH sensor exists that uses a hydrogel co-
polymer film deposited over interdigitated electrodes (IDEs), and the film?s water 
content changes depending on the  presence of acids or bases, resulting in a pH-
dependent level of hydrogel swelling and ultimately measured by the change in 
electrode surface conductivity [79]. By exploiting these material properties, it is 
possible to design sensors with signal outputs that change proportionally to the 
environment and specifically to a target of interest [80]. On that note, biomaterials-
based films can act as enzyme-specific degradable substrates, which I leveraged when 
deposited over our system?s capacitive sensors [72], [73], [76], [78]. Because the GI 
19 
 
 
 
tract is a host to numerous hydrolytic enzymes, many of which are summarized in Table 
1-1, monitoring them in situ would likely offer pathologically relevant information. 
Table 1-1 Major gastrointestinal enzymes, organized by source and excluding the salivary glands and 
large intestine, and some of their respective associated health conditions. 
Source Enzyme Substrate Conditions 
Stomach Pepsin Proteins Carcinoma, pernicious anemia, 
post-operative peptic and 
duodenal ulcers [81] 
Lipase Triglycerides Pancreatic dysfunction, cystic 
fibrosis [81] 
Pancreas Amylase Starch Pancreatitis (Acute/Chronic), 
Lipase Triglycerides Adenocarcinoma, Pancreatic 
Proteases (Trypsin, Peptides insufficiency [1], [2], [65], [82] 
Chymotrypsin) 
Elastase Elastin 
Nucleases Nucleic Acids 
Phospholipase Phospholipids Adenocarcinoma [83]?[85] 
Carboxypeptidase Protein terminal 
amino acid 
Small Enterokinase Trypsinogen, Mucosal villous atrophy, 
Intestine Chymotrypsinogen Hypoproteinemia, Pancreatic 
Insufficiency, Carcinoma [86], 
[87] 
Maltase Maltose Mucosal villous atrophy, 
Coeliac?s Disease [87], [88], 
Lactase Lactose Lactose Intolerance, Intestinal 
hypermotility, Diarrhea, 
Inflammatory bowel disease [89], 
Sucrase Sucrose [90], Sprue [91] 
 
 The above enzymes, it is worth noting, are those present in higher 
concentrations in the GI environment and critical to homeostatic physiological 
function. As mentioned in the previous section, other enzymes have been isolated from 
pancreatic secretions that become present in pathological conditions such as pancreatic 
cancer. One example is matrix metalloproteinase (MMP), or matrixin, which is a type 
of protease that can be subdivided into more than 20 different forms [92]. Specifically, 
MMP-9, or gelatinase B, has been present in both serum and pancreatic juices, and aids 
20 
 
 
 
in differentiating patients with pancreatic ductal adenocarcinoma, the most common 
form of pancreatic cancer, from those with chronic pancreatitis [56], [93]. MMP-9 
explicitly hydrolyzes peptide sequences with Pro-X-X-Hy-(Ser/Thr) motifs at P3 
through P2 positions, so it is possible to design hydrolytic substrates that would react 
similarly to the aforementioned materials [93]. One example has been demonstrated by 
Biela et al., where the peptide has been used to crosslink Dextran films, and the 
degradation was monitored via EIS [94]. Their hydrogel fabrication process, which 
could be utilized for other peptide-hydrogel-crosslinking mechanisms, is illustrated in 
Figure 1-7 below. While the process for forming peptide-crosslinked hydrogels has 
been previously demonstrated, the above work showcases a use-case with a clinical 
target. Using EIS or other electronics sensing methods like those discussed in Section 
1.3.1, such film-based platforms have the potential for integration with 
microelectronics that be implemented in capsule devices for measuring these targets in 
the GI tract. 
(a)  (b)  
Figure 1-7 (a) Process for fabricating peptide-crosslinked hydrogels over gold IDE sensors, specifically 
for sensing MMP-9 with EIS, and (b) 10 Hz impedance response of resulting sensor to different MMP-9 
concentrations. Reproduced with permission from [95]. 
21 
 
 
 
1.3.4 Integrated In situ Capsule Technology 
 This section introduces the system-based solution that is pursued in this 
dissertation. Integrated systems embedded with automated micro- and 
nanotechnologies are gaining momentum as revolutionary methods in biomedical 
applications for replacing cumbersome or complex processes. Many of these systems, 
like most automated technology, require a combination of sensors or various electronic 
components, as well as some form of interface with human physiology to achieve a 
designated purpose. 
Several such devices, depicted in Figure 1-8, are continuously under 
development for vastly different applications, and are ubiquitous both in industrial and 
academic settings. One of the most successful commercially available examples is the 
capsule endoscope, such as the PillCam by Medtronic, introduced as an alternative to 
endoscopic procedures in the GI tract such as a colonoscopy or EGD [10]. Capsule 
endoscopes are essentially pill-shaped devices containing a camera that, upon 
ingestion, allows for tetherless video streaming and recording of the GI tract to locate 
abnormalities such as ulcers, sources of bleeding, polyps, or tumors. While there is also 
an associated non-trivial risk of bowel obstruction, the primary benefit is that it is able 
not only to reach areas that exceed the range of other procedures, but it also does not 
require anesthesia or as much physician involvement [96].  
22 
 
 
 
 
 
Figure 1-8 Examples of integrated capsule systems. From left to right: PillCam COLON 2, GI gas sensor, 
and GI bleeding monitor. Reproduced with permission from [97]?[99]. 
Additional examples in academia use specific sensing mechanisms when the 
target becomes more complex. One device exists for measuring gas concentrations in 
the gut [100]. Here, CO2, H2, and CH4 are measured with electrochemical sensors at 
various sampling sites through the stomach, small intestine, cecum, and colon, and are 
measured in response to controlled diets based on fiber content. Though 
electrochemical gas sensors have been around since the 19th century, the novelty lies 
primarily with the integration of this sensor into an ingestible system [101]. This 
becomes compounded by evidence suggesting gas composition throughout the GI tract 
changes in various pathological conditions, thereby offering potential for clinical 
evaluation. Another device uses fluorescence imaging to determine the occurrence of 
GI bleeding [99]. By injecting fluorescein into the bloodstream, the marker naturally 
makes its way into the GI tract, where a capsule housing a fluorimeter detects and 
wirelessly transmits data to an external monitoring unit. This device serves to enhance 
traditional endoscope procedures for real-time monitoring, and is also more specific 
than using capsule endoscopy as discussed above.  
Though there are additional similar devices that warrant discussion, a prevalent 
theme is the integration of multiple modules. The modules included often consist of a 
23 
 
 
 
sensor, microcontroller, signal transceiver, power supply and management electronics, 
and a capsule-like package, enabling the development of devices capable of targeted, 
real-time sensing with wireless signal transmission. A brief list of recent state-of-the-
art capsule devices are presented in Table 1-2. 
 
Table 1-2 Integrated capsule systems. From left to right: PillCam COLON 2, GI gas sensor, and GI 
bleeding monitor. 
Product/ Transducer/Target Communication Power Manufacturer/ 
Application Reference 
Gas Sensing Electrochemical/ 433 MHz Silver Oxide  [12], [13], 
CO2, O2, H2, CH4 [100] 
PillCam/ Optical 434 MHz/ Lithium Ion Medtronic/ 
MiroCam/ 430 MHz/ IntroMedic 
Endocapsule 434 MHz Co./ 
Olympus 
GI Bleeding Photodetectors/ 2.4 GHz (Zigbee) Lithium [99] 
Fluorescein polymer 
GI Bleeding Photodetectors/ 433 MHz Lithium [102] 
Bacterial light Manganese 
GI Electrochemical 433 MHz Lithium [103] 
Electrochemical Manganese 
Signatures Dioxide  
pH/ Capacitive/ 434 MHz Silver Oxide [104] 
Temperature/ Resistive 
Pressure 
 
 The above devices are but a short list of the emerging technologies in the 
ingestible capsule domain, and are representative of the open-ended nature of 
integrated capsule system development. Though the most consistent feature is the ISM 
band of radiofrequency (RF) communication for wireless signal transmission (433-434 
MHz), the transducers and applications vary significantly and each require their own 
signal-conversion circuit to be compatible with the microcontroller capable of tasks 
such as (a) generating an input signal at programmable intervals, (b) signal processing, 
and (c) enabling feedback control for modulating power or signal characteristics, 
24 
 
 
 
among others. One aspect of note with Table 1-2 is that the optical capsules are the 
only ones listed with manufacturers. Though there are additional capsules on the market 
for sensing features such as temperature or pressure, those used for capsule endoscopy 
remain the most mature and popular for commercial development and therefore 
medical diagnostics.  
 In terms of power supply, or battery, each capsule must identify the most 
appropriate option to ensure the on-system electronics can maintain individual modular 
functions. First and foremost, batteries must operate over the time intervals required, 
which is often throughout the entirety of the GI tract and can average at 34 hours with 
a range of 8-57 hours [105]. This is generally easier for systems designated to operate 
for shorter time durations, i.e. when localized to specific organs or organ transitions. 
However, this is critically dependent on the electronics of the system for achieving its 
respective function. Each option in Table 1-2 is adequate for the devices presented, 
while there is likely room for optimization through considerations such as data 
compression (of the transmitted signal) or power-saving algorithms that alter the power 
modes of each module; for example, some IC components can operate between modes 
such as ?Active? or ?Deep Sleep? (specifically for the MCU discussed in Chapter 3), 
which correspond to current consumption when wireless functions are triggered or 
delayed, respectively. In capsule endoscopy, for example, power consumption of the 
signal transmitters has evolved with each iteration, ranging between 2-7 mW, and is 
detailed further in other works and for other types of ingestible devices [106]?[108].  
One of the most important features of these devices is the packaging, which 
upon observation appears relatively consistent and is the only barrier between the 
25 
 
 
 
internal system and the external environment. The primary metric the design packaging 
must achieve is safety for both the GI environment, i.e. the tissue, fluids, etc., and for 
the internal components. The former is dictated by the shape, dimension, materials, and 
sealing capability of the device shell. The shape is almost always a capsule, where the 
length in one axis is longer than the length of the other, or the width, and the ends of 
the shell on the longer axis are rounded [109]. Work from Baek et al. investigated the 
effect of structural features such as geometry, length, and diameter on capsule friction, 
producing downstream expectations on how it would impact GI transit [110]. Though 
capsules have long been a standard for ingestible technology in the area of 
pharmaceuticals, their findings yielded adequate rationale to appreciate the impact of 
this seemingly trivial feature. Furthermore, the ability to control transit time poses 
significant utility for additional applications. For example, in the event that a device 
would be required to remain in a specific GI region for a longer time scale, one could 
use geometric innovation such as a greater surface area at the micro- and nanoscale to 
increase frictional forces. While such strategies may augment additional complications 
such as biofouling or potential trauma to GI mucosa if large enough to create 
perforations, it adds further consideration to ingestible capsule design [110]?[114].  
In terms of materials, the most common requirement is biocompatibility. 
Though biocompatibility can still entail a wide range of properties, the most popular 
use in this context is for the material to be tested as a ?surface device that contacts 
breached or compromised surfaces for prolonged contact?, according to the Food and 
Drug Administration (FDA)-modified ISO-10993 [115]. This would entail testing such 
as cytotoxicity, sensitization, irritation or intracutaneous reactivity, acute systemic 
26 
 
 
 
toxicity, material-mediated pyrogenicity, subacute/subchronic toxicity, and 
implantation, as outlined in Table A.1: Biocompatibility Evaluation Endpoints [115]. 
The effects prove challenging to enable efficient control due to the heterogeneity of GI 
fluids, even without considerations for variation in diet. For example, the pH of the GI 
tract can range from <1 (highly acidic) to >7 (neutral, slightly basic), imbuing a 
requirement with a significant impact because (1) this comprises approximately half 
the pH range in our general chemical understanding and (2) this limits the number of 
materials that would not react with such pH levels. This is compounded by numerous 
other species in GI fluids likely to induce any number of physicochemical interactions 
with the capsule package, such as proteins (enzymes, glycoproteins, etc.), hormones, 
ions, etc. [116].  
The materials must also withstand mechanical forces occurring during 
peristalsis, which can vary depending on the region of the GI tract. In the small 
intestine, the primary forces acting on devices in transit are traction force, contact force, 
and peristaltic waves [117], [118]. Forces are slightly different in the stomach, where 
tonic contractions and peristaltic waves dominate the mechanical impact on materials 
to induce movement toward the pyloric sphincter; these forces were also investigated 
using pills with embedded pressure sensor arrays [119], [120]. For the above capsules, 
cladding and sealing materials vary where mentioned. Arefin et al. only discusses 
having used corning sealant while using a 3D printed cladding to surround their system 
[121]. The work by Mimee et al. uses PDMS as their sealing material without any 
cladding material [102], and the capsule for GI bleeding uses a machined cylinder made 
of acetal resin, specifically Delrin (DuPont, DE). The optical capsules generally use a 
27 
 
 
 
variety of biocompatible plastics such as polycarbonate, polyurethane, or nylon due to 
their resistance to gastric acids [122], [123]. Chapter 3 will revisit the concept of 
capsule packaging, including the materials I utilized for enhanced utility for both 
protecting our system and targeting our GI region of interest. 
1.4 Structure of Dissertation 
Chapter 1 introduced the reader to the motivation and background required to 
appreciate the impact of this dissertation, ranging from the rationale of our sensor and 
system design to the requirements for the downstream development of a platform 
adaptable to a variety of healthcare diagnostics. Chapter 2 delves into the intricacies of 
impedance sensing and its utility for garnering detailed information on surface-
interacting hydrolytic reaction mechanisms. Chapter 3 presents our packaging-focused 
approach in hybridizing 3D-printing and pharmaceutical polymer coatings into a 
cohesive manufacturing strategy. Chapter 4 discusses the challenges associated with 
assembling, testing, and validating our system, while demonstrating its versatility for 
sensing multiple duodenal biomarkers. Finally, Chapter 5 summarizes the contributions 
of the work while introducing various directions for expanding device utility. 
 
 
28 
 
 
 
2. Chapter 2: Electrical Transduction of Pancreatic Trypsin 
Activity 
2.1 Pancreatic Enzyme Sensing 
As discussed in Table 1-1, GI enzymes can be categorized by their region of 
origin, substrate, and associated pathologies. Their quantification is traditionally 
performed through assays, including various forms of enzyme-linked immunosorbent 
assays (ELISA), with a range of methodologies including colorimetric, radioimmuno, 
and immunochemiluminometric. What remains consistent among each of them is the 
requirement of transducers to convert the biological signal, i.e. the enzyme 
concentration or activity, to a measurable signal for analysis in electronic systems. As 
discussed in Section 1.3.1, electrical biosensing techniques leverage direct changes in 
electrical properties of the substrate and biomarker-substrate interface, while achieving 
inherently more simplicity than those discussed above [124]. Such direct measurements 
can be useful not only for detecting the biomarker of interest, but determining 
qualitative or mechanistic details on how the sensor materials behave in response to 
their activity; this can aid further in improving sensor designs or optimizing materials. 
Here, I describe efforts to observe changes in electrical characteristics of substrate films 
after exposure to hydrolytic enzymes. 
Pancreatic enzymes, including trypsin, lipase, and amylase, are excellent model 
biomarkers to assess GI health and proper pancreatic function.  Pancreatic trypsin, in 
particular, is a protease essential to a functional digestive system, allowing for the 
breakdown of various proteins to enable nutritional absorption in the small intestine. It 
is secreted as the proenzyme trypsinogen by acinar cells, where it travels through the 
29 
 
 
 
pancreatic duct into the duodenum and becomes activated upon hydrolysis by 
enterokinase on the lumen surface [7]. Lipase and ?-amylase are among the other major 
enzymes that are activated in the duodenum. These enzymes enter the duodenum in a 
basic solution with the intent to neutralize the gastric acid entering from the stomach. 
In various types of pancreatic pathologies, such as chronic pancreatitis or pancreatic 
adenocarcinoma, trypsin activity has been shown to decrease, indicating the acinus or 
duct has been compromised in some way [1], [125], [126]. 
 
Figure 2-1 Schematic of collagen denaturation into gelatin. Reproduced with permission from [127]. 
Recent techniques for sensing trypsin involve analyzing substrate degradation 
in response to enzymolysis, described in Section 1.3.3. Trypsin hydrolyzes specific 
peptide bonds present on substrates with amino acids lysine and arginine. The 
substrates include cytochrome c, gelatin, albumin or poly-L-lysine, while transduction 
methods are equally variable, using either optical, fluorescence, ionic conductance, or 
resistance [71], [73], [128]?[130]. Because of its versatility for processing and 
30 
 
 
 
modification, including over other electrode systems, gelatin was investigated as a 
model substrate for this work [131], [132]. Gelatin is a hydrolytic product of collagen, 
and the 10%?20% of most final gelatin forms consist of the amino acids mentioned 
above  [133]. Gelatin is also biocompatible, and therefore an attractive substrate 
material for trypsin. As a hydrogel that begins as a low viscosity liquid when heated, it 
can be formed to a variety of structures that are limited only by the tools used until it 
cools and transitions from sol to gel state, the process for which is depicted in Figure 
2-1.  
One such structure is a thin film, compatible for various impedimetric or 
amperometric sensors made from standard monolithic fabrication. Thin film structures 
have been achieved through deposition of hydrogel, while in the sol state, using 
strategies such as spin-coating, drop-casting, dip-coating, etc. [134]. Furthermore, 
crosslinking is utilized often to enhance the stability of the material, and is limited by 
the reactive functional groups in the structure?s chemistry [135]. Glutaraldehyde, for 
example, has been used to allow for gelatin films to maintain structural integrity at 
higher temperatures ? i.e. the physiological temperatures present throughout the GI 
tract (37-39?C) ? to ensure the film remains stable, though high concentrations can 
induce toxic effects on cells [136], [137].  
It is important here to introduce several distinctions between the sensors 
intended for the system in this thesis, i.e. in vivo in the GI tract, and how most 
biochemical sensors for biomedical applications operate. Firstly, in vitro systems are 
often not concerned with toxicity as they are often utilized in vitro, where the samples 
are either in buffer or fluids that have been removed from the cellular contact. This 
31 
 
 
 
does not apply to the few sensing strategies that are designed to operate in vivo, as there 
are significant risks associated when implementing foreign objects in the physiological 
environment due to the potential for an immune response with the complex biochemical 
heterogeneity [138]. This introduces another distinction, which is that in vivo sensing 
has limited capabilities regarding sample pretreatment without additional system 
modules, and thus the sensor is subject to interact with each molecule ? including those 
that are not of interest and therefore nonspecific ? present in the fluid. Therefore, when 
targeting trypsin, which, as discussed above, originates near the duodenum, it is critical 
to test the sensor with fluids exhibiting similar heterogeneity to thus reduce 
unprecedented signals resulting from these nonspecific interactions.  
Through our use of gelatin, there are various potential nonspecific interactions 
at risk. For example, though none of the species are likely to react with the substrate 
via hydrolysis ? except for chymotrypsin, as discussed in Section 1.3.3 ? other 
interactions may occur such as ionic or molecular diffusion into the film or aggregation 
at the film surface. Diffusion often occurs for species with sizes smaller than the 
hydrogel film mesh, which can be mitigated through tighter crosslinking through 
techniques discussed above. Aggregation, on the other hand, can occur when the 
species are larger than the mesh or are immobilized via nonspecific binding at the film 
surface [139]. Both can yield distinct sensor responses, depending on the transducer 
used, though I focused in this work on testing several species that fit into the latter 
category. 
In this chapter, I discuss our efforts for developing a sensing strategy for 
measuring trypsin activity in solutions containing several types of interfering enzymes 
32 
 
 
 
characteristic of duodenal fluid. Crosslinked gelatin films were formed over 
interdigitated finger electrode (IDE) sensors, which were fabricated similarly to those 
described previously [140]?[142]. These sensors were inserted into fluidic chambers 
and exposed to various pancreatic enzymes while changes in the mass and impedance 
responses were recorded. Ultimately, it was possible to differentiate the presence of 
trypsin and provide insight as to its role in producing complex impedimetric signals, 
offering potential for utilizing this method for in situ sampling. 
Dr. Luke Beardslee contributed to fluidic setup design and sensor process 
development. Dr. Young Wook Kim contributed to initial sensor design in Appendix 
A.1. Dr. Sowmya Subramanian and Dr. Thomas Winkler contributed respectively to 
initial data extraction (Appendix C of [143]) and analysis of impedance results. This 
work was published in contribution to [144]. 
2.2 System Design and Experimental Setup 
2.2.1 IDE Sensor Fabrication and Film Deposition  
A simplified schematic of the sensor fabrication process is depicted in Figure 
2-2a. A pair of sensor electrodes per device (Appendix A.1), one sensor with substrate 
polymer coating and one for negative control (footprint 3 cm ? 1 cm), was obtained 
using a standard photolithography process for NR9-1500PY photoresist (Futurrex, Inc., 
Franklin, NJ) over 100 mm pyrex wafers. This involved a 5-minute dehydration period, 
spin-coating at 3000 RPM for 30 seconds, a 150?C soft bake for 1 minute, exposure to 
365 nm UV 190 mJ/cm2, a 100?C hard bake for 1 minute, development in RD6 for 10-
12 seconds, then rinsing in DI water. Cr/Au (20 nm/200 nm) was deposited via E-beam 
33 
 
 
 
evaporation(Angstrom Engineering, Ontario, Canada), followed by liftoff with acetone 
for 5 minutes. The resulting IDE fingers, a total of 146, were 2 ?m wide with 4 ?m 
spacing and 960 ?m in length, as shown in Figure 2-2b.  
Next, a 75 ?m thick polyvinyl chloride (PVC) film (Semiconductor Equipment 
Corp, Moorpark, CA, USA) was applied to the wafer surface, and patterns were 
manually cut over one sensor per device as a mask for the gelatin film. Gelatin type B 
solution (from bovine skin, Sigma Aldrich, St. Louis, MO, USA) was added to 50 ?C 
deionized (DI) water while stirring until complete dissolution to a final 5 %w/v. 200 
?L gelatin solution was spin-coated at 400 rpm for 30 s over the sensors, then cooled 
overnight into a gel state over the test sensor. All gelatin films were crosslinked after 
overnight cooling with 0.5% glutaraldehyde solution in DI water at room temperature 
for at least 4 hours to improve thermal stability. After PVC removal from the IDE 
patterned wafer, gelatin film thicknesses were characterized, or profiled, with a contact 
profilometer (Veeco, Dektak-8 stylus, Plainview, NY, USA) which were found ranging 
between 400?700 nm. IDE devices were then diced using a standard dicing saw 
(Microautomation, Centreville, VA, USA), rinsed with DI water, and dried overnight. 
(a) (b)  
Figure 2-2 (a) Schematic of sensor fabrication. Step 1 entailed photolithography and E-beam deposition 
of Cr/Au for 20nm/200nm, then liftoff with acetone. After overlaying a 75 ?m thick PVC film, 5% gelatin 
in buffer was spin-coated at 400 rpm for 30 sec, cooled overnight, then crosslinked with 0.5% 
glutaraldehyde solution to improve thermal stability. After PVC removal, gelatin was measured using a 
contact profilometer at thicknesses of 300-700 nm. (b) Microscopic image of the resulting sensors. 
34 
 
 
 
2.2.2 Enzyme Solution Preparation  
Trypsin from bovine pancreas (?7500 benzoyl-L-arginine ethyl ester units/mg 
solid, 23.8 kDa), ?-Amylase (?10 units/mg solid, 51?54 kDa) and lipase (Type II, 100?
500 units/mg protein, 48 kDa) from porcine pancreas, as well as phosphate-buffered 
saline (PBS) capsules were obtained from Sigma Aldrich. 0.1 M PBS (pH 7.4) was 
used in each experiment. Enzymes were mixed and incubated in PBS for at least 15 
min at 37 ?C before use. Trypsin solutions of either 1 or 0.5 mg/mL were used, while 
lipase and ?-amylase were prepared at 1 mg/mL each. Heterogeneous solutions 
consisted of 1 mg/mL of each enzyme. 
2.2.3 Impedance Sensing 
To test the sensors with continuous enzymes, a fluidic chamber was designed, 
depicted in Figure 2-3, that enabled electrical connections with the sensor contact pads. 
Diced IDE sensors were inserted into a custom two-piece 3D printed chambers 
containing an inlet and outlet for respective solution entry and removal from the 
chamber via continuous flow. The parts were designed in SolidWorks (Dassault 
Systemes, V?lizy-Villacoublay, France) and 3D printed from MED610 with an Objet30 
Pro (Stratasys, Eden Prairie, MN, USA). Fluid was injected with syringe pumps at 20 
?L/min through 0.4/2.24 mm inner/outer diameter tygon tubing (Cole-Parmer, Vernon 
Hills, IL, USA) into the chamber inlet, then exited through the outlet into a waste 
beaker. Sensor response was recorded with each medium until the signal saturated at a 
constant magnitude, beginning with air, PBS (negative control), then PBS with 
enzymes, described in Section 2.2.2. The chamber was incubated at ~37 ?C for each 
experiment to simulate physiological temperature. IDE contact pads were connected to 
35 
 
 
 
a CHI 660D Electrochemical workstation (CH Instruments, Bee Cave, TX, USA), one 
pad connected to the working terminal and the other pad to shorted reference and 
counter terminals. The frequency was swept from 10 Hz?1 MHz at an AC voltage 
amplitude of 50 mV for two minutes per sensor. Sensor substrates were weighed with 
a Sartorius ME-5 microbalance (Sartorius, Goettingen, Germany) and profiled, through 
the process described in Section 2.2.1, at three separate time-points: (1) before 
deposition, (2) once the films were dehydrated, and (3) following experimentation to 
correspond absolute change of film mass and thickness with the resulting changes in 
impedance. 
 
(a)  
(b)  
Figure 2-3 (a) Fluidic reaction chamber for testing sensors with enzyme solutions. (b) Schematic of the 
film-coated sensor and experimental chamber (left). Photograph of Experimental setup (right). 
36 
 
 
 
2.2.4 QCM Sensing Methodology 
As a means of measuring activity in real time between the enzymes discussed 
above and the gelatin films, I investigated mass sensing using quartz crystal 
microbalance (QCM) (Inficon, East Syracuse, NY, USA). Fluid was injected with 
syringe pumps at 20 ?L/min through tygon tubing (Cole-Parmer) into the QCM 
Teflon? 100 ?L chamber inlet, then exited through the outlet into a waste beaker. 
Sensor response was recorded with each medium until signal saturation, beginning with 
air, PBS (negative control), then PBS with enzymes, described in Section 2.2.2. The 
chamber was incubated at ~37?C for every experiment to simulate physiological 
temperature. QCM substrates were cleaned before deposition and after experimentation 
with 3:1 Piranha solution. They were also weighed with a microbalance and profiled as 
described in Section 2.2.1. 
2.2.5 Image/Data Analysis 
Drop-cast deposited and crosslinked over each dye, with one of the two sensors 
consistently covered with PVC. This process was done carefully to cover only one of 
the sensors on the device as to maintain the presence of a control sensor. The resulting 
films possessed thicknesses ranging from 60?90 ?m (Veeco). Films were observed and 
analyzed using an INM100 Microscope (Leica, Wetzlar, Germany) and an S-3400 
Variable Pressure Scanning Electron Microscope (SEM) (Hitachi, Tokyo, Japan). 
Impedance data were exported from CHI software as .txt files, and imported, along 
with film thickness profiles, into MATLAB (MathWorks, Natick, MA, USA) for 
analysis. Nyquist and Bode spectra were observed, and impedance at various 
frequencies were plotted over time for each condition.  
37 
 
 
 
2.3 Results and Discussion 
2.3.1 Film Structure and Morphology 
Microscopic images of resulting IDE sensors, with and without spin-coated 
gelatin films before exposure to enzyme solutions, are displayed in Figure 2-4a. The 
gelatin films, depicted over the sensors on the left side of the image, were slightly 
frayed at the edges after removal of the PVC tape; however, this was not found to 
directly affect sensor response as the edges were further than a finger width distance 
from the IDEs. Figure 2-4b?d represent the dried sensor after exposure to either PBS 
(b), trypsin (c), or a combination of trypsin, lipase, and amylase (d) in PBS. After 
surface profiling, it was evident that trypsin alone removes the most gelatin from the 
sensor surface, visualized in Figure 2-4c, while each other condition leaves a residue 
of film over the test sensor and more non-specific adhesion of other enzymes to the 
control surface, such as the mixture-treated sensors depicted in Figure 2-4d. Table 2-1 
describes the mean proportion of remaining film thickness after exposure to each 
solution condition, calculated using equation (3), where t is the thickness at each 
measurement point explained above: 
???? ? ????? (3) 
???? ????????? ????????? =  ( ) 
????
A value of 1 indicated zero film loss, and each thickness was achieved after signal 
saturation and complete drying of the sample in an ambient environment. 
38 
 
 
 
 
Figure 2-4 Resulting interdigitated finger electrode (IDE) sensors, with (Left) and without (Right) gelatin 
films, (a) pre-exposure to solutions, and post-exposure to (b) buffer; (c) trypsin; or (d) all three enzymes 
in buffer. Scale bar = 1 mm. 
 
Table 2-1 Ratio of spin-coated film remaining after various enzyme conditions, analyzed with a contact 
profiler (n = 2). 
Condition Film Thickness Remaining (%) 
PBS (Phosphate Buffered Saline) 78.6 ? 0.05 
Amylase (1 mg/mL) 24.3 ? 0.08 
Lipase (1 mg/mL) 52.4 ? 0.30 
Trypsin (1 mg/mL) 11.1 ? 0.03 
Trypsin (0.5 mg/mL) 21.7 ? 0.10 
 
Scanning electron microscope (SEM) images of the drop-cast films in various 
conditions can be viewed in Figure 2-5, where (a-b) are side-view and (c-d) are top-
down. In Figure 2-5a, the dry, cross-linked films resulted in an edge thickness of ~60 
?m. After exposure to trypsin for several hours, at which the absolute impedance had 
saturated, the sensor was rinsed with DIH2O and completely dried. The resulting edge 
thickness, depicted in Figure 2-5b, was then measured at ~9 ?m, reflecting the removal 
of film as a result of hydrolysis from trypsin. The ratio of remaining to original film 
39 
 
 
 
thickness was slightly less than 0.15, which is relatively consistent with the ratio of 
film remaining with the spin-coated film from Table 2-1 (0.11). Furthermore, the film 
is much smoother, and can be observed again from the top view in Figure 2-5c. 
  
(a) (b) 
  
(c) (d) 
Figure 2-5 Representative SEM images of drop-cast, crosslinked gelatin films. Edge views of film (a) 
without exposure to enzyme solutions and (b) after exposure to trypsin, and top views of film (c) after 
exposure to trypsin and (d) after exposure to all three enzymes in buffer. 
After the drop-cast film was exposed to a combination of all three enzymes, the 
thickness reduced to only ~0.68 of the original film after the impedance signal had 
saturated. This indicates saturation occurred before even half of the film had been 
degraded. Further, as depicted in Figure 2-5d, globules form in the film, some of which 
can exceed 15 ?m in diameter. This is likely due to the entrapment of non-specific 
enzymes in the gelatin matrix, potentially responsible for attenuating the trypsin 
activity [145]?[147]. Further, the immobilization of the non-specific enzymes in the 
gelatin may induce tighter bonds (i.e., aldimine with the amylase) to form, possibly 
40 
 
 
 
explaining the decrease in thickness[148]. However, based on the results in Table 2-1, 
there is still less degradation with all three enzymes than what occurs with only amylase 
or lipase, suggesting enzymatic cross-interference in the solution prior to interactions 
occurring with the gelatin surface. Due to the prevalence of arginine and lysine in the 
amino acid compositions of both amylase and lipase, trypsin may be performing 
additional proteolysis on the nonspecific enzymes in the solution as a result, hence the 
necessity for further experimentation in the future [149], [150]. 
2.3.2 Impedance Sensing of Trypsin 
I characterized the impedance response of gelatin film degradation over IDE 
sensors in response to various stimuli in a fabricated impedance testing setup made 
from 3D-printed components, PEEK tubing, and a potentiostat, discussed in Section 
2.2.3. Impedance spectra were recorded after film stabilization under PBS flow with 
the 3D-printed setup in Figure 2-3. The absolute impedance recorded at 10 kHz 
(indicated later in this chapter in Figure 2-7b) revealed the highest sensitivity to the 
presence of trypsin during flow due to the largest changes that occurred relative to the 
initial signal saturation. Figure 2-6 presents impedance spectra comparing the effect of 
different concentrations of trypsin, which show that 0.5 mg/mL trypsin reduced the 
film ?? rate to 0.114 k?/min, a significant drop compared to the 2.88 k?/min rate of 
the 1 mg/mL concentration. This indicates that the ?? rate can potentially be used a 
sensing parameter. The curve for ?Trypsin?No Film? essentially represents the 
impedance of PBS over the sensor since trypsin is not significantly polar and thus does 
not alter the dielectric properties of the buffer, while ?Control?Film? represents the 
impedance of the film after equilibrating with the PBS. 
41 
 
 
 
 
 
Figure 2-6 Representative impedance responses at 10 kHz of sensors, either uncoated or coated with a 
gelatin film, to trypsin at 1 (both film and no film), 0.5 (film only) and 0 mg/mL in phosphate-buffered 
saline (Control). ?Trypsin?No Film? represents the impedance of PBS over the sensor, and ?Control?
Film? represents the impedance of the film after equilibrating with the PBS. The error bars plot the 
temporal change in impedance at respective time points (temporal span = 3, n=3). 
Interestingly, the impedance of the film alone is expected to be higher than that 
of PBS due to the gelatin?s relatively neutral isoelectric point [151]. It is likely, 
however, that equilibration with the PBS flow allowed salt ions to accumulate 
compared to the non-PBS exposed films (data not shown), thereby increasing the ion 
concentration in the gelatin matrix over the sensor via diffusion.  
In Figure 2-7, I report a representative sequence of Bode plots at different 
exposure times that convey changes in impedance with respect to the equivalent 
electrical circuit model of the film, such as the double-layer capacitance and solution 
resistance. 
42 
 
 
 
  
(a) (b) 
  
(c) (d) 
Figure 2-7 Representative Bode and Nyquist plots for progressive exposure of (a-c) uncoated and (b-d) 
gelatin-coated sensors to 1 mg/mL trypsin in buffer. Continuous (-) and dotted (--) lines indicate |Z| and 
the phase, respectively. (n=3). 
Here, I observed over time, both the phase and impedance spectra of the coated 
sensor approaching those of the uncoated sensor, indicating a decrease in film (though 
it cannot be objectively determined as a decrease in coverage or thickness based on this 
data) that reflects trypsin degradation of the film due to increased exposure of the 
sensors to the buffer. I also observed in Figure 2-7b that with increasing frequency the 
phase approached 90?, further evidence that resistive elements become more negligible 
compared to capacitive elements in the circuit. Additionally, there is an increase in the 
solution resistance (RS) in the Nyquist plot, illustrated by the increase in diameter of 
the semicircles (Figure 2-7d) as trypsin continues to remove film from the sensor 
surface. This is different compared to what is observed in uncoated sensors (Figure 
43 
 
 
 
2-7c), which further supports the likelihood of ion accumulation in the film. Our 
observations indicate that each of these circuit elements, i.e. resistance and capacitance, 
can be monitored to determine film degradation, which suggested that I could simplify 
our electrochemical sensing strategy to measure a specific signal, therefore reducing 
signal and power requirements to adapt to a smaller electronics system. This first 
requires investigation as to potential non-specific interactions to determine if they 
affect either parameter, as well as determine which would be ideal for enhanced 
sensitivity to the film reactions. 
2.3.3 Impedance Response with Enzyme Mixture 
 After characterizing the impact of trypsin on the impedance of the film-coated 
sensors, I then observed the impact of introducing non-specific enzymes to the 
solution. In Figure 2-8, I observed that the addition of both amylase and lipase (1 
mg/mL each) to trypsin caused a significant decrease in rate of ??, measured at 6.77 
? 10?4 k?/min compared to 2.88 k?/min for the trypsin only. Further, when amylase 
or lipase are alone in buffer without trypsin, I observed a decreasing impedance, 
opposite of the result found with buffers including trypsin. The origin of the increase 
in impedance during the first hour of the sensors exposed to lipase, as well as within 
two hours for the amylase, still require investigation, but only seem temporary as the 
long-term impact of the enzymes yields a decrease in impedance.  
 
44 
 
 
 
 
Figure 2-8 Representative impedance responses at 10 kHz of sensors, either uncoated or coated with a 
gelatin film, to trypsin, amylase, or lipase, each at 1 mg/mL, or combinations of each enzyme with trypsin. 
The error bars plot the temporal change in impedance at respective time points (temporal span = 3, n=3). 
 
Upon analyzing the Nyquist plots of non-specific enzyme impact in Figure 2-9, 
I observed additional trends, again different to those from trypsin, induced by the 
addition of amylase and lipase. These trends include a decrease in the radius of the 
semicircular region (representative of the RS), the latter of which can be represented by 
equation (4): 
?
??  =  ?  (4) ?
where ? is the solution resistivity, l electrode length, A is the electrode area [152]. This 
effect on the RS can be a result of the lack of film degradation, or even increase in film 
thickness [153]. The semi-circular region in the film responding to amylase or lipase is 
much more defined, indicating a larger double-layer capacitance of the active material 
i.e. the film and its constituents. This characteristic of the film is also maintained while 
there is buffer alone, suggesting the non-specific enzymes do not significantly affect 
the structure of the film in the absence of trypsin (indicated from Figure 2-5d). 
45 
 
 
 
Alternatively, in the case of most nonspecific enzymes, residual glutaraldehyde 
remaining in the film may have allowed for accumulation of additional ?-amylase or 
lipase, likely observed in Section 2.3.1 [145]?[147]. Upon measuring the film after 
exposure to all three enzymes, I see the similar trend of increasing impedance as when 
trypsin is alone, though at a much slower pace. I again observed the phenomena of 
decreasing RS in the Nyquist plot, similarly to what occurred with the amylase and 
lipase alone. The reduction in degradation rate discussed above may be alleviated by 
modifying the film to prevent non-specific binding, such as with the use alternative 
crosslinkers (microbial transglutaminase has shown to be effective), inhibitors (orlistat 
or tendamistat for lipase and amylase, respectively) or with more specific film materials 
to trypsin (i.e., poly-l-lysine) [154]?[156]. Overall, however, the experiments 
conducted above allowed us to observe the impacts of these enzymes on the impedance 
of gelatin films, and it appeared that the presence of trypsin, rather than the non-specific 
enzymes, is responsible for inducing increments in impedance over the gelatin-coated 
sensors. 
 
46 
 
 
 
(a)   
(b)  
(c)  
Figure 2-9 Representative Nyquist plots for gelatin-coated sensors to (a) 1 mg/mL lipase; (b) 1 mg/mL 
amylase; and (c) trypsin, amylase, and lipase, each at 1 mg/mL over time. (n=3). 
47 
 
 
 
2.3.4 QCM Results 
Figure 2-10 displays the QCM response of gelatin films when exposed to 
different solutions; PBS, different trypsin concentrations (0.5 and 1 mg/mL 
concentrations), amylase or lipase at 1 mg/mL concentration, or combinations (each at 
1 mg/mL) of trypsin, amylase, and lipase. First, the mass response equilibrated under 
air for approximately 20 min (not displayed), and further equilibrated under the initial 
flow of PBS solution without enzymes for approximately two hours. Figure 2-10a 
compares the response to different concentrations of trypsin in buffer. The average 
gelatin degradation rates, calculated using the region with the largest slope magnitude, 
that occurred for 1 mg/mL and 0.5 mg/mL trypsin concentrations were 0.52 and 0.16 
mg/min, respectively, while there was no significant degradation that occurred for 
lipase, amylase, or PBS alone, as indicated in Figure 2-10b. For the single enzyme 
mixtures in Figure 2-10c, gelatin degradation rates were quantified at a maximum of 
0.09 and 0.12 mg/min for mixtures with lipase and amylase, respectively. For all three 
enzymes (each 1 mg/mL), the average rate decreased further to 0.04 mg/min. Since not 
every film possessed the same initial mass, the total change in mass is presented, with 
most of the film having been removed from the substrate by the 1 mg/mL trypsin. 
Additionally, values were recorded until the signal saturated, as saturation time varied 
depending on the amount of film removed from the substrate for each trypsin 
experiment. Though the slope for all three enzymes in Figure 2-10c combined appears 
steeper than those for just trypsin and either amylase or lipase individually, the time 
scale gives an indication that the rate of film removal from the substrate decreased 
significantly as the signal approached saturation. This indicates amylase and lipase 
48 
 
 
 
interfere with the gelatin removal from the substrate, though it is unknown as to 
whether they affect trypsin activity or the availability of peptide bonds in the gelatin. 
(a)  
(b)  
(c)  
Figure 2-10 Representative QCM response of gelatin film to (a) trypsin at 1, 0.5 and 0 mg/mL in PBS, (b) 
trypsin, amylase, or lipase, each at 1 mg/mL, or (c) combinations of each enzyme with trypsin (each at 1 
mg/mL). (n=3). 
49 
 
 
 
 This section provided empirical evidence as to the feasibility of using gelatin 
films for measuring pancreatic trypsin through electrical impedance spectroscopy. I 
measured various electrical properties of the films such as overall impedance, which is 
then decomposed using Nyquist analysis and QCM to compare isolated trypsin versus 
nonspecific enzyme results. 
2.4 Film Reactions to Varying Enzyme Concentrations 
 Because I observed real-time changes in electrical properties of the gelatin films 
in response to pancreatic enzymes and a potential dependence on physical properties 
such as thickness, I wanted to further evaluate limitations of how these films react. In 
this section, I exposed the films developed in the previous section to a wider 
concentration range of pancreatic trypsin and observe potential time-scale 
dependencies in the film thickness and mass response. 
 Gelatin films were deposited via drop-casting of 500 ?L of 1 %w/v gelatin in 
DIH2O over pre-weighed 10mm x 10mm Pyrex chips, then refrigerated overnight and 
crosslinked similar to the process described in Section 2.2.1. The mass and surface 
profiles were then obtained for each film-coated chip (FC). 
 FCs were immersed in glass petri dishes containing solutions, under 200 RPM 
stir via magnetic stir bar and 37-39?C, consisting of 0.1 M PBS only or PBS with 
varying concentrations of trypsin (1 mg/mL, 100 ?g/mL, 10 ?g/mL, and 1 ?g/mL). FCs 
were then removed at 30 minutes, 1 hour, 2 hour, and 4 hour time points. After 
removing 3 samples at each time point, FCs were then rinsed with DIH2O, blow-dried 
with N2, then set aside to dry in ambient for 24 hours. FCs were then re-weighed and 
re-profiled. 
50 
 
 
 
 Figure 2-11 illustrates the trends produced for both % change in mass and 
thickness, using equation (3) and its converted form when considering the change in 
mass. Initial film masses and thicknesses were calculated at 0.572?0.01 mg and 
6.80?0.44 ?m thick, respectively. The thickness for the drop-cast films are likely lower 
than those from Section 2.2.1 due to the volume used and the available surface area and 
how they impacted potential spreading effects. I observed distinct changes in the rate 
of mass and thickness decrease with increasing trypsin concentration. The time points 
chosen only yielded saturation levels for the 100 ?g/mL and 1 mg/mL concentrations, 
which appear to be evident at 2 hours of incubation, though it is clear the higher 
concentration induced faster degradation of the film. 
(a)  
(b)  
Figure 2-11 Gelatin film responses to varying trypsin concentrations through analysis of resulting 
%change in in (a) mass and (b) thickness, with each sample removed from incubation at different time 
points. Error bars depict standard deviation (n=3). 
51 
 
 
 
 
 I then used these trends to calculate the rate of reaction over time with each 
concentration, depicted in Figure 2-12 below. The time frames used for 1, 10, 100, and 
1000 ?g/mL were, for mass, 2, 0.5, 0.5, and 0.5 hours, respectively, and for thickness, 
2, 1, 0.5, and 0.5, respectively, due to the length of time observed until saturation. Not 
only do I observed that the changes in values occur in logarithmic trends, but that they 
are almost equivalent to each other for both mass (mg/hr) and thickness (?m/hr). This 
is critical in relating these metrics to enzyme concentration, which are quantified using 
the depicted equations and their respective R2 coefficients for variance. 
 
Figure 2-12 Logarithmic rate of reactions of gelatin films in solutions with varying pancreatic trypsin 
concentrations, calculated using the slopes in percent change of film mass and thickness. Error bars depict 
standard deviation (n=3). 
 In this section, I briefly described our efforts to characterize the reaction 
response of gelatin films to varying concentrations of pancreatic trypsin. I evaluated 
the impact specifically on film mass and thickness, and enable us to use these metrics 
as means to evaluate the substrate reaction as a subsequent result from the presence of 
environmental trypsin. 
52 
 
 
 
2.5 System Modeling for Capacitive Sensing 
2.5.1 Electrical Behavior of Biomaterial Films  
Based on the impedance response demonstrated in Figure 2-6, I observed that 
the capacitive and resistive elements of the equivalent circuit model of the IDEs both 
change significantly as the gelatin film hydrolyzes from the trypsin. While the total 
impedance shifted more significantly at lower frequencies (<100 kHz) reflecting more 
capacitive effects compared to other frequency range, I also noticed measurable shifts 
at higher frequencies for resistive effects (>100 kHz), indicating the potential to adapt 
sensing strategies for either signal type. Because impedance sensing possesses more 
circuit requirements, adding further challenge to miniaturization of the sensing 
electronics, I was interested in focusing on transitioning from impedance sensing 
strategy to solely either a capacitive- or resistive-sensing strategy and determining 
which would have more sensitivity to changes in film reactions. 
Capacitive sensing possesses several benefits over resistive sensing in 
understanding how thin films structures over IDEs affect either. When a thin film is 
deposited onto an electronic transducer, the double-layer capacitance (Cdl), is affected, 
which can be expressed as the sum of the constant capacitance from an unmodified 
electrode (CAu) and a surface modifier (Cmod), depicted in Figure 2-13 and expressed in 
equation (5) [39]. Cmod is an indicator of binding to the electrode surface, dependent on 
the relative permittivity (?r) and thickness of the film (t), and can be calculated by 
equation (6), where ?0 is vacuum permittivity (8.85x10
-12 F/m) and A is the electrode 
area [153]. Equation (7), solving for CAu, is similar, only with film thickness replaced 
by the distance between fingers (d) [157].  
53 
 
 
 
(a) (b)  
Figure 2-13 (a) Equivalent circuit model and (b) schematic for thin films over impedance sensors. 
Reproduced with permission from [39]. 
1 1 1
= +  (5) 
??? ??? ????
???0?
???? =  (6) ?
???0?
??? =   (7) ?
The benefit of measuring capacitance lies specifically in Cmod. To clarify, 
resistive sensing of a degrading insulating film would require that the thickness of the 
film be reduced at its entirety at one specific location along the surface area for any 
charge-transfer to occur. This would also affect the Cdl at the location, so the response 
time would be equivalent. However, Cmod is affected by changes in the rest of the film 
thickness whereas RS, described in Equation (4), is not, thereby indicating that the 
sensor would respond upon initial degradation at a certain change in thickness that 
would occur before the reaction has reached the sensor surface. While changes in RS 
were evident through the Nyquist plots from impedance measurements, Equation (6) 
indicate measuring the capacitance alone can induce a faster change in signal. 
Ultimately, the more the film is degraded, the more CS and RS are prominent in their 
effect. In Appendix B, a COMSOL simulation of the equivalent circuit model is 
presented, which further highlights the feasibility for capacitive sensing. 
54 
 
 
 
2.5.3 Sensor Characterization with Varying Film Thickness 
In determining the geometries for the next iteration of sensors used in 
measuring film degradation in response to enzymatic activity, I developed an optical 
mask, depicted in Figure 6-2 of Appendix A.2, with varying widths and spacings of the 
IDE electrodes. This was critical for understanding the impact of different film 
thicknesses on the resulting capacitance and how sensor performance could be 
optimized with different geometries. Each die on the mask also consisted of four 
sensors for both (a) higher throughput sensing of single targets or (b) sensing of 
multiple targets with each sensor output acquired through different multiplexer (MUX) 
inputs. This will be discussed further in Chapter 3. 
Fabrication of the sensors was conducted via drop-casting as described in 
Section 2.2.1. This was because I was also interested in seeing how the film profile and 
thickness consistency would change when achieved at the individual die level versus 
the complete wafer level, as the latter had produced significant variability in the 
coverage of the film over the wafer, resulting in differing film thicknesses for devices 
at the center compared to those at the outer wafer perimeter. Additionally, I wanted to 
establish the expected film thickness when reusing the sensors with newly deposited 
films. After dicing, the resulting 4-sensor devices, where dicing tape remained over the 
contact pads as a mask, placed on  the spin coater, 200 ?L gelatin solution (5 %w/v in 
DIH2O) was loaded on top with pipette, and the spinner was set to spin for 30 seconds 
at varying spin speeds. After spinning, the sensors were cooled overnight in ambient, 
crosslinked, and measured with a profilometer similar to those in Section 2.2.1. 
55 
 
 
 
 (a)  (b)  
(c)  
Figure 2-14 (a) Resulting gelatin film thicknesses and different spin speeds, both pre- and post- rinsing 
with DIH2O. (b) Top-level view of example sensor outline, with each sensor label. (c) Resulting film 
thicknesses for each spin speed, broken down by sensor ? using labels from (b) ? and rinsing. Error bars 
depict standard deviation (n=3). 
Figure 2-14a depicts the resulting film thicknesses when sensors were spin-
coated with 2 mL gelatin solution at 400, 600, and 800 RPM, as well as comparing 
them before and after rinsing with DIH2O and drying. There is an evident decrease in 
thickness with increasing spin speeds which follows the typical trends found in spin 
coating recipes, though it must be noted that these are all both (a) thicker than total 
wafer spinning at 400 RPM (400-700 nm) and (b) thinner than drop-casting alone (60-
90 ?m), as described in Section 2.2.1. A closer look, however, using the top view sensor 
image in Figure 2-14b, indicates that there is still lateral thickness variability in Figure 
2-14c, where I observed that the films remain thicker for the two central sensors (b & 
56 
 
 
 
c) but are thinner for the outer sensors (a & d), likely due to surface tension of the liquid 
to the wafer. Fortunately, this variability does appear to decrease with increasing spin 
speed, as evident primarily with the 800 RPM spun sensors after rinsing, indicating a 
possible better drying efficiency. 
 The sensors were then analyzed for their series capacitance (Cs) and resistance 
(Rs) using an LCR meter (Agilent E4980A) controlled with LabVIEW. The values were 
calculated using the circuit model depicted in Figure 2-15a. First, depicted in Figure 
2-15, I compared how each response changed both in the presence and absence of 
gelatin films, the latter of which were those spun at 800 RPM and after rinsing. Here, I 
first noticed that Cs and Rs increases and decreases, respectively, with decreasing finger 
spacing. This is expected through calculating Cs with equation (7) ? using conditions 
Cs = CAu when there is no film (Cs-no film), and Cs = Cdl when there is a film (Cs-film) ? 
where CAu  ? d (finger distance) and was calculated using, ??=1.005 for air, and A = 
4.4x10-6 m2 (interfacing area, constant over different dimensions used in this work). 
For Rs, however, there is no reported dependence on the film thickness, and the trend I 
observed between the "Film? and ?No Film? conditions did not aid clarity to the 
relationship.  However, the measured Cs-no film varied compared to CAu, which when 
taking the ratio of the measured value over the calculated value was referred to as an 
offset factor that is likely an effect of parasitic capacitances. Using the resulting Cs-film, 
I calculate the dielectric constant of the gelatin film (????) using equation (8), as A and 
?? remain constant: 
???????????
???? =   (8) ?????????
 
57 
 
 
 
This yielded an ???? of 1.23?0.06, which acts as a relatively efficient insulator (likely 
considering it had little to no moisture) and I could therefore use ? along with their 
respective film thicknesses ? to measure each respective Cmod, using equation (6), 
verifying there was a dependence of capacitance on the film thickness. I then used these 
factors to calculate the Cdl using equation (5). These results are summarized in Table 
2-2. 
 (a)  
6
5
4
25
3
50
75
2
1
0
No Film Film
(b)  
6
5
4
25
3
50
75
2
1
0
(c) No Film Film  
Figure 2-15 (a) Series equivalent circuit mode of LCR meter. (b-c) Electrical (capacitive and resistive, 
respectively) analysis of sensors across varying electrode spacings and comparing between the presence 
and absence of a film. Error bars depict standard deviation (n=3). 
58 
 
 
Rs (k?) Cs (pF)
 
Table 2-2 Parameters used for capacitance analysis of gelatin films. 
CS ? No Film 
CAu (pF) 
Width/Spacing (pF) Offset Factor 
(Calculated) 
(?m) (Measured) 
75 0.587 1.653?0.35 2.8 
50 0.779 2.073?0.27 2.7 
25 1.56 2.8?0.77 1.8 
Film Thickness CS ? Film (pF) ???? CMOD (pF) CDL (pF) 
(?m) (Measured) (Calculated) (Calculated) (Calculated) 
1.27?0.71 1.99?0.1 1.205 36.93 1.582 
1.15?0.81 2.45?0.06 1.182 40.04 1.971 
1.04?0.8 3.62?0.1 1.294 48.43 2.650 
 Mean 1.23?0.06   
 
Further analysis, displayed in Figure 2-16, was done to observe the effect of 
sensor spacing on sensitivity to the change in thickness. Regression analysis was 
performed with different thicknesses measured, and I found, using the slope from each 
linear fit trend, that the capacitance increased with increasing thickness more 
significantly when the sensor finger spacing decreased, indicated by the depicted 
formulas. I also measured resistance, which was also affected by the film thickness, but 
in the opposite direction ? i.e. the sensitivity decreased with reduced IDE spacing ? 
while the resistance overall would decrease as the film thickness increased. However, 
regression on these slopes indicated that this change in sensitivity for varying finger 
spacing was also greater for capacitance (slope=1.0?10-3) than resistance 
(slope=8.0?10-4), indicating that the response would be more sensitive by extracting 
sensor capacitance. 
59 
 
 
 
(a)  
(b)  
Figure 2-16 Regression analysis of different gelatin film thicknesses and their impacts on sensor (a) 
capacitance and (b) resistance across 25, 50, and 75 ?m electrode spacings. 
 
60 
 
 
 
 While these results provide some insight as to the utility of measuring the 
electrical properties of these gelatin films, there are various limitations as to the extent 
of this utility. For example, because the films were dried hydrogels, the responses will 
vary significantly depending on their hydration level and respective conductivity when 
high ion or electrolyte concentrations in the surrounding media diffuse into the films; 
the former is limited with higher cross-linking, thus preventing it as a major concern, 
though a correction from a humidity measurement may also improve accuracy. This 
may impact film thickness as well, however sensitivity of these particular sensors was 
only tested within a select range of film thicknesses, preventing us from ascertaining 
to what extent the sensitivity changes at lower and higher thickness ranges, especially 
when thinner than the width of the electrode fingers. Additionally, it would be 
beneficial to test these responses when there was additional media, such as buffer, 
present above the films, which may affect the resulting double layer capacitance and 
solution resistance. Finally, even though I presented in Section 2.4 that the hydrolysis 
reaction between trypsin and gelatin affects the thickness ? as well as the mass ? of the 
film, not all reactant-substrate interactions manifest in such surface effects as opposed 
to the bulk or at more isolated regions. This will be addressed more in chapter 4. 
 In this section, I described efforts to relate film thickness to electrical properties 
using literature-based mathematical and equivalent circuit models, as well as perform 
basic computational modeling to establish relationships between film thickness to 
sensor capacitance. I further developed strategies for deposition of gelatin films onto 
microfabricated IDE electrodes, where sensor characterization with different electrode 
geometries is used to relate the resulting film thicknesses to intrinsic electrical 
61 
 
 
 
properties of the films ? i.e. capacitance and resistance. This will be used toward 
determining parameter requirements in designing a microelectronics-based measuring 
system for film-based biosensing. 
2.6 Conclusions 
The work in this chapter demonstrated an impedance-based sensing strategy for 
measuring film-degradation using gelatin as a substrate for pancreatic trypsin. While 
sensor fabrication proved relatively straightforward, the information obtained offered 
broad insight to enzymatic interactions that occur between the film and various species 
in the media. I observed a linear relation for the rate of change in impedance over time 
by measuring the degradation rate of gelatin films by trypsin, and that nonspecific 
enzymes diminish this effect. The changes in impedance, as well as other electrical 
properties as determined in Section 2.5, resulting from changes in film morphology and 
thickness, are likely to translate to similar film and enzyme-targeting combinations. 
Decomposing the impedance properties into its fundamental components 
provides various implications for system design and development into a capsule-based 
microsystem. In isolating capacitive and resistive elements of the film response to 
biochemical reactions, I evaluated their respective efficacies to determine optimal 
choices in the target system metric for assessing the local presence of enzymes. 
Minimizing system elements to the necessary components is critical when developing 
a microelectronics system, aiding in subsequent reduction in form factor to scales 
relevant for different domains, such as the human body. Potential strategies ? i.e. a 
capacitive sensing microelectronic system ? that could be implemented when 
62 
 
 
 
measuring biomaterial film reactions to biomolecular targets in complex environments 
such as pancreatic secretions in vivo will be discussed further in the following chapters. 
63 
 
 
 
3. Chapter 3: Packaging and System Miniaturization for 
Gastrointestinal Navigation 
 
3.1 Capsules and Gastrointestinal Sampling 
As discussed in Section 2.1, various techniques are used for measuring GI 
enzyme content. However, they generally require multiple steps such as sample 
preparation and multiple rinsing cycles, preventing them from use for in situ sensing 
[158]?[161]. More importantly, samples must first be extracted using other tools and 
processed externally. Duodenal contents, for example, are traditionally extracted via 
endoscopic aspiration then analyzed through assay kits, while various sensing protocols 
for more specific analytes require separation techniques such as centrifugation or 
filtration [1], [47], [162]. For more direct sampling of GI fluidic contents, protocols 
vary between merely collecting stool contents to endoscopic aspiration, depending on 
the target [163], [164]. 
To transition to in situ biosensing for the benefit of real time data acquisition 
and procedural simplification as described in Chapter 1, strategies must be 
implemented that can either (1) reduce the number of steps involved or (2) incorporate 
them into the sensing system for rapid detection upon sampling. Toward this end, 
capsule-based integrated devices have recently been gaining traction as vehicles for 
sampling through the gastrointestinal (GI) tract that have the potential to achieve these 
tasks and overcome various limitations in current medical techniques, such as location 
restrictions or insufficient sensitivity measures [165], [166]. To become a viable 
medical solution for its respective function, capsules in the form of autonomous 
diagnostic systems must be able to navigate and function within the GI tract?s 
64 
 
 
 
physiological environment, which is often chaotic and inhibitive to systems requiring  
control over the features affecting the sensing interface, which includes but is not 
limited to fouling or non-specific interactions [99], [100], [103], [167]?[170]. With the 
wide range of potential applications, capsules are becoming more ubiquitous in 
academic settings, and a few are experiencing success in industry aiding practitioners 
in the clinical field [171]?[173]. As mentioned previously, one of the most successful 
commercially available examples of capsule-based integrated devices is the capsule 
endoscope, such as the PillCam (Medtronic). It was initially introduced as a solution to 
extend the limited range of endoscopic procedures in the GI tract such as a colonoscopy 
or esophagogastroduodenoscopy [10], [174], [175]. The primary benefit of this 
technology over traditional endoscopy, other than that it does not require anesthesia or 
as much involvement from the clinician, is its enhanced ability to access difficult-to-
reach GI regions. Overall, these detection technologies aim to understand the state of 
the GI environment for identifying pathological conditions or obtaining previously 
inaccessible information in a noninvasive approach.  
GI fluids are used in diagnostics for isolating biomarkers related to specific GI 
pathologies, and targeting specific regions for sampling is critical for associating the 
biomarker with its source. Additional examples that have yet to mature to clinical 
implementation require specific sensors that often include complex surface 
modifications for binding or reacting with a target species, based on the biomarker of 
interest. One such device measures gas concentrations in the gut, which optical images 
cannot easily distinguish [100]. Though electrochemical gas sensors have been around 
since the 19th century, the novelty lies primarily with the integration of this sensor into 
65 
 
 
 
an ingestible system, thereby offering clinical potential that did not previously exist 
[101]. Another device uses fluorescence imaging to determine the occurrence of GI 
bleeding [99]. This device serves as another alternative to traditional endoscope 
procedures for real-time monitoring, with more specificity than using capsule 
endoscopy as discussed above. Additional examples warrant further discussion over 
the impact of capsule-based biomedical devices, and a prevalent theme is the 
integration of low-power electronics and relevant sensors within a compact ingestible 
package [11], [103], [168]?[170], [172]. 
The most important feature of capsule-based technologies is their departure 
from benchtop analytical systems toward miniaturization into an in vivo sensing 
platform. For example, while traditional endoscopy possesses more functions than 
capsule endoscopy, the latter isolates the task of image capture and transmission and 
delegates the required subtasks to various modules and components within the capsule 
package. The basic modules involved are depicted in Figure 3-1 below [176]. 
 
Figure 3-1 Modules contained within endoscopy capsule: (1) Optical Dome, (2) LED, (3) Short-focus lens, 
(4) CMOS image sensor, (5) RF model, (6) MCU, and (7) Power model [176]. 
The basic components required for most in vivo real-time sensing with wireless 
signal transmission include: a sensor, microcontroller, signal transceiver, power 
66 
 
 
 
supply, and a capsule-like package that can be swallowed [177]. However, for 
achieving alternative tasks, such as sensing, at the required size scale, energy remains 
a significant limiting factor that demands controlled allocation of power to as few 
system modules as possible for minimal time durations. Furthermore, complex tracking 
systems are often required to monitor whether or not the sensor measurements are 
occurring in the appropriate region [178]?[181]. These systems can include 
cumbersome equipment that the patient must wear in the form of a belt-like sensor 
array unit, and rely on magnetic or RF-based localization technologies that possess 
significant power requirements [182], [183]. While knowing the location of the capsule 
is necessary for certain functions where targeting requires more precision, such as 
monitoring transit time at specific regions or isolating positions of tumors or bleeding, 
the technologies described above still retain several other fundamental limitations. For 
one system referred to as the MTS-1 (magnetic tracking system), intestinal and 
abdominal movement can affect measurement accuracy, requiring simultaneous usage 
of computer tomography, and therefore additional costly equipment, to obtain 
anatomical data [178].  
Though there are clear benefits to these anatomical tracking systems, they are 
not always necessary for the application of interest. Other than capsule endoscopes, 
they are rarely implemented in capsules designated for measuring specific biomarkers. 
Because these biomarkers are generally secreted at specific locations in the GI tract, 
we can leverage various features of these regions toward ensuring that the system can 
target them. One such feature that varies consistently throughout the GI tract is the pH, 
as illustrated in Figure 3-2 [184], [185]. The pH balance is critical to healthy GI 
67 
 
 
 
function, as imbalances can be indicative of or lead to harmful conditions such as 
inflammatory bowel disease (IBD) or irritable bowel syndrome (IBS), among others 
[186]?[189]. The pH is also necessary for regulating enzyme and other digestive 
functions, as most enzymes, such as those relevant to this work in Table 3-1, possess 
optimal pH ranges [190], [191]. Using materials that respond directly to pH can aid as 
a passive means of not only targeting the appropriate region and its respective fluid, 
but ensuring the fluid has a pH within the range for optimal enzyme activity, such as 
those discussed previously in this thesis. 
 
Figure 3-2 Human GI tract morphology with respective pH ranges. Reproduced with permission from 
[184]. 
Table 3-1 Various GI enzymes and their pH for optimal activity [16]. 
Enzyme pH Optimum 
Gastric Lipase 4.0-5.0 
Pepsin 1.5-1.6 
Pancreatic Lipase 8.0 
Trypsin 7.8-8.7 
Pancreatic Amylase 6.7-7.0 
Urease 7.0 
Maltase 6.1-6.8 
 
68 
 
 
 
In this chapter, I elaborate on our development of a wireless electronics system 
and its encapsulation in a capsule package for monitoring the ability to sample fluid at 
various targeted GI regions. The electronics interface with similar sensors as those 
developed in chapter 2, but that (a) require low power and (b) use surrounding pH to 
approximate location. A 3-D printed custom package encapsulates capacitive sensors, 
which are multiplexed and interfaced via a capacitive-voltage converter (CVC) with a 
Bluetooth system-in-package (SiP) that can pair with a mobile device via Bluetooth 
Low Energy (BLE). The capsule design possesses gratings that are plugged with 
different polymers, which dissolve under various biochemical conditions to allow for 
selective fluid entry into a sensing chamber. Once the environmental condition (pH in 
this study) is met, the dissolution of the polymer in the grating is detected by the 
capacitive sensors due to fluid entry changing the dielectric property of the chamber.  
The polymers tested were chosen to demonstrate potential for sampling in a 
variety of regions in the GI tract based on pH, with an emphasis on sampling pancreatic 
secretions as depicted in Figure 3-3. The transition in pH at the duodenum from acidic 
to neutral conditions due to reactions of gastric juices with pancreatic bicarbonate, 
which occurs over a time period of ~30-60 minutes, offers a distinct event and time 
frame that is used in our design to target and sample pancreatic secretions. The gratings 
are filled with various formulations of pH-sensitive copolymers composed of acrylic 
and methacrylic acid that are currently used in the pharmaceutical industry for drug 
delivery to specific GI regions [192]?[194]. Different coating thicknesses are tested to 
evaluate control over dissolution at varying time scales, as well as combinations, or 
69 
 
 
 
layers, of coatings for more complex sequences. Here, I discuss the design of the system 
packaging, electronics, and their miniaturization. 
Dr. Luke Beardslee contributed to the circuit and PCB design for the electronics 
system, as well as some of the initial capsule concepts. Justin Stine contributed to 
electronics design, programming, and troubleshooting. Rajendra Mayavan Sathyam 
contributed to microcontroller programming and development of the android 
application for signal acquisition. This work was published in contribution to [195]. 
 
Figure 3-3 Depiction of system outlook and application (Image credits: National Cancer Institute). 
Polymer coatings are intended to dissolve at either the stomach (pH 1.5-3) or the duodenum (pH 7-8.5). 
At the duodenum, gastric acid-neutralizing bicarbonate is secreted from the pancreas via the sphincter 
of Oddi. 
3.2 System Description 
3.2.1 Sensors and External System Electronics 
Electrodes consisting of four sensors with a footprint of 3 cm x 1 cm (Appendix 
A.2) were obtained through a standard ?lift-off? process as described in Section 2.2.1. 
70 
 
 
 
The resulting interdigitated electrodes (IDEs), a total of 72 pairs, were 50 ?m wide with 
50 ?m spacing and 2 mm long. IDE devices were then diced using a standard dicing 
saw (Microautomation, Centreville, VA). 
The electronics, including various off-the-shelf integrated circuits (IC), used to 
operate the capsule are summarized in Figure 3-4. In our configuration, four identical 
capacitive transducers are connected with a multiplexer as inputs (MUX IC, DG4052, 
Analog Microelectronics, Mainz, Germany), each of which is measured using a CVC 
IC (CAV444, Analog Microelectronics). The voltage output is connected to the analog-
to-digital converter (ADC), a part of the microcontroller unit (MCU, BGM121, Silicon 
Labs, Austin, TX), which transmits the signal either serially to a computer via UART 
or wirelessly to a mobile device via BLE, respectively. General-purpose input/output 
(GPIO) pins from the MCU control the MUX to determine which sensor connects to 
the CVC. BLE is used as the wireless signal transmission technology because of the 
ease of interfacing with a smartphone, the availability of components that fit the needed 
form factor, and the availability of low-power components with a sleep mode. The 
rechargeable 14 mAh at 3.7 V lithium polymer battery (Powerstream, Orem, UT) was 
the optimal choice for the design; it satisfies the constraint of a small form factor and 
is able to source enough power for wireless transmission. Figure 3-5 presents the 
calibration of the ADC output potential using capacitors with known values. In 
addition, this figure displays data from the sensors being inserted into and removed 
from buffer via UART and BLE, respectively, where the slight elevation in signal after 
removal is due to residue buffer remaining on the sensor surface. 
71 
 
 
 
 
Figure 3-4 System electronics overview. Circuit design includes a multiplexer (MUX) integrated circuit 
(IC), a capacitance-voltage-converter (CVC) IC, a 1.5 MHz step-up DC-DC converter, and a BLE system-
in-package (SiP) micro-controller unit (MCU). 
 (a) (b)  
Figure 3-5 (a) Circuit calibration with known capacitors across four multiplexed inputs. The resulting 
system senses capacitive changes in the 0.8-220 pF dynamic range with a sensitivity of 3.2x10-3 pF/mV 
and operated using a 3.3 V source. (b) Data comparison between Bluetooth versus serial (UART) 
communication in different experiments, where the capsule (open gratings) was measured in air, inserted 
into buffer, then removed. Error bars depict standard deviation (n=4). 
3.2.2 Capsule Package Design and Materials 
Each capsule is designed in SolidWorks (Dassault Systemes, V?lizy-
Villacoublay, France) and 3-D printed from a biocompatible acrylic resin (MED610) 
using an Objet30 Pro (Stratasys, Rehovot, Israel) to assess the feasibility and utility of 
the capsule packaging and embedded gratings for sampling, as well as achieve the 
desired form factor according to previously determined standards for capsule 
endoscopy [10], [174]. The capsule designs and prints in Figure 3-6 are used for 
72 
 
 
 
experiments using an external circuit layout, where only the sensors are inserted into 
the capsule and the electronics remain external on a soldered protoboard (Adafruit, 
New York, NY). Gratings consisting of a 3x3 array of 0.4 mm x 0.4 mm holes, which 
function primarily as chamber inlets, were positioned over sensor regions for fluid 
accessibility into a hollow space, or sensing chamber. A slot with 600 ?m thickness 
and 1.1 cm width was created to allow for sensor insertion while sensor contact pads 
remained on the outside of the capsule. The sensor and slot interface was epoxied for 
sealing, then the capsule was immersed for 1 second in highly viscous 14% Eudragit S 
100 (Evonik, Essen, Germany), a methacrylic acid?methyl methacrylate copolymer 
with 0.3% sodium laurylsulfate in methanol for dip-coating individual layers over up 
to 5 successive cycles. No bubbles appeared, indicating the polymer solution was not 
displacing air in the sensing chamber and therefore did not cause the capsule to fill and 
contact the sensors. Cross-sectioned images were taken of the grating regions, two of 
which (0 and 1 coating) are depicted in Figure 3-7a-b, and analyzed for thickness over 
grating and non-grating regions in Figure 3-7c. 
 
(a)  (b)  
Figure 3-6 (a) CAD model of the capsule used with external circuit. (b) 3-D printed version with zoomed 
up view of gratings. Scale bar = 1 mm. 
73 
 
 
 
(a)  (b)  
(c)  
Figure 3-7 Cross-sectional optical images of the sampling gratings (a) uncoated and (b) with 1 coating of 
polymer, where yellow arrows indicate the grating locations and yellow boxes indicate the inlets. (c) 
Characterization of polymer coating thickness vs. number of dipping steps from both non-grating and 
grating regions over the 3-D-printed capsule surface. Error bars depict standard deviation (n=3). Scale 
bar = 1 mm. 
3.2.3 System Miniaturization and Assembly 
The circuit was embedded into a four-layer printed circuit board (PCB, 25 mm 
x 10 mm x 2 mm) made of FR4 with a silver finish (Sunstone, Mulino, OR). The 
PCB, assembled by Screaming Circuits (Canby, OR), is depicted in Figure 3-8a, 
indicating the pins used to connect to the battery, the contact pads for connecting to 
the sensors, and the pins for programming the MCU. The schematic of the assembled 
capsule setup is depicted in Figure 3-8b. New sensors (Appendix A.3), with a smaller 
footprint to fit between the PCB contact pads while maintaining the same capacitance 
range, depicted in Figure 3-8c, were fabricated as described previously. The new 
74 
 
 
 
sensor IDEs were 10 ?m wide and 1.5 mm length, while the new finger spacing of 10 
?m increased the sensitivity from 3.6 (for the 50 ?m spacing sensors) to 7.3 pF/mV. 
These sensors were designed to be utilized with the biomaterial films that were 
described in Section 2.4. Most experiments were performed with the PCB connected 
to a power supply through wires, as the battery performance had not been optimized 
at this point for experiments requiring long durations. The PCB and wires were 
encapsulated in a conformal coating, consisting of acrylic copolymer (CAIG 
Laboratories, Inc., Poway, CA) for circuit and electrical insulation, and inserted into a 
modified 3-D printed capsule which directly interfaced the chamber inlets with the 
sensors. A photograph of the assembled capsule is depicted in Figure 3-8d, while the 
system specifications are listed in  
Table 3-2. 
 
75 
 
 
 
 
Figure 3-8 (a) PCB rendering with sensor die fixated between designated pads using epoxy. Contact was 
made by curing conductive silver ink. (b) Schematic of the assembled capsule setup. (c) Calibration of 
BLE MCU at the PCB level using standard known capacitors. (d) Photograph of assembled capsule. Scale 
bar = 5 mm unless otherwise specified. 
 
Table 3-2 System Specifications 
3D-Printed Shell  Electronics 
Outer Diameter 12.7mm  PCB Length 25.8 mm 
Inner Diameter  11.5 mm  PCB Width 10.4 mm 
Length 35 mm  Sensor Capacitance Range 0.8-220 pF 
Inlet area 4 mm2  Sensor Capacitance Sensitivity 7.3 pF/mV 
Sensors  Operating Voltage Operating Voltage 
Finger Width 5 ?m  Current Consumption Active: 5 mA 
Finger Spacing 5 ?m   Deep Sleep: 2.5 ?A 
Finger Length  750 ?m  Battery (Powerstream) Li-Polymer/14 mAh/3.3 V 
Number of Fingers 80  Wireless Communication BLE 2.4 GHz 
Finger Thickness  200/20 Au/Cr nm    
 
3.2.4 Experimental Procedures for pH Sequences 
For the circuit layout external to the capsule, the sensor contact pads were 
connected to the circuit while sandwiched between PCBs with soldered spring-loaded 
76 
 
 
 
pins. The capacitance was then transmitted serially from the BGM121 wireless starter 
kit (WSTK) via UART. All capsules were inserted into heated solutions (37?C) under 
stirring conditions (250 RPM). A measurement sequence was defined to evaluate 
polymer coating degradation at various pH states, according to a known pH dissolution 
profile. Beginning with measurements in air for 5 min, the capsule was then inserted 
into a solution with a negative control pH, the level of which depended on the polymer 
plug used. The pH was then adjusted progressively to the positive control pH that was 
expected to cause dissolution. Solutions beginning with acidic pH consisted of 0.1 M 
acetic acid (pH 2.9), while solutions beginning with neutral pH were 0.1 M phosphate 
buffer (pH 7.6). These pH levels were chosen to reflect the pH range of the target 
regions of the GI tract, specifically the stomach and small intestine for acidic and 
neutral secretions, respectively [196], [197]. Progressive pH adjustments were 
performed every 30 min for acid- and neutral-solutions via droplets of 1 M phosphoric 
acid and 1 M sodium hydroxide, respectively, measured using a standard pH meter 
(Oyster-10, Extech Industries, Nashua, NH). The coating procedure discussed in 
Section 3.2.2 was repeated, three coatings each, for Eudragit formulations L 100 and E 
PO, intended to dissolve at > pH 6 and < pH 5, respectively. The pH progression for 
their respective experiments were 2.9 to 7.6 and 7.6 to 2.9 in discrete increments of 
~1.0 to control for the sensor responses while beginning in non-soluble pH conditions 
designed to inhibit dissolution. The procedure was repeated with 3 coatings of L 100, 
followed by 3 coatings of E PO (depicted in the coating rendering in Figure 3-10d), 
and tested with a pH sequence of 7.6 to 2.9 to 7.6. Once the positive control pH was 
77 
 
 
 
reached, the capsule was removed from solution for 10 min and signal transmission 
was terminated. 
The miniaturized system of the capsule with inserted PCB was coated with three 
coatings of E PO solution, then immersed into both pH 7.6 and pH 2.9 solutions while 
recording capacitance over time. The recorded capacitance was obtained via BLE by 
pairing with an Android phone, initiated via a modified application from the 
manufacturer (Silicon Labs). The data was stored onto the mobile device as a text file 
and analyzed using a desktop PC. Photographs were taken at the 0, 30-minute, and 1 
hour time points during immersion to record progress in coating dissolution. All 
polymer solutions were 30%w/v in methanol and dyed for enhanced visualization.  
All data was plotted using a custom MATLAB (MathWorks) graphical user 
interface (GUI). For the WSTK obtained data, the signals were recorded into a text file 
using RealTerm. For the PCB layout, the recorded capacitance was obtained via BLE 
by pairing with an Android phone, as described previously. 
3.3 Experimental Results 
3.3.1 Sensor Responses 
The sensor responses, both in voltage at the ADC and the corresponding 
capacitance, for capsules with varying numbers of coatings are depicted in Figure 3-9. 
For 0 coatings, the point at which the capsule was inserted into buffer is indicated by 
the sharp increase in capacitance at around 5 minutes due to immediate entry of buffer, 
causing an instant change in the medium dielectric constant at the sensor surface. The 
data for 1 coating was omitted due to the lack of sealing of the gratings, producing a 
78 
 
 
 
similar result as that of the 0-coating sequence. With an increase in the number of 
coating layers, the rate of the capacitance change is reduced significantly, recorded at 
approximately 16.0, 5.6, and 1.1 pF/min for 0, 2, and 3 coatings, respectively. The 
system with 4 coatings did not open within the 140-minute duration, though some drift 
in capacitance due to a potential increase in humidity was seen. As expected, additional 
dip coatings corresponded to an increase in the fluid entry time, allowing us to tailor 
the packaging to different time scales based on coating thickness from corresponding 
measurements in Figure 3-7c. 
 
Figure 3-9 Sensor responses (inserted into buffer at 5 min) of capsules with different numbers of S 100 
coatings (0, 2, 3, and 4, respectively) with increasing pH using titers of 1 M NaOH. Error bars depict 
standard deviation (n=4). Chamber filling is characterized by a capacitance change of ~50 pF from the 
initial capacitance in air. For each formulation, it was determined that 3 coatings were optimal for the 
w/v% used. 
3.3.2 Coating Variety and Combinations 
After varying the coating thickness, I tested polymer formulations that would 
dissolve at different pH ranges. Real-time capacitance measurements for the pH 
sequences of the L 100 and E PO coated capsules (3 dip-coatings each) are depicted in 
79 
 
 
 
Figure 3-10a-b, respectively, with the latter pH in which the coatings were intended to 
dissolve enough to achieve sampling; the distinct drop in pH toward the end of each 
sequence denotes when the capsule was removed from the sample. The resulting 
capacitance indicated that sampling occurred in the appropriate ranges (neutral for L 
100 and acidic for E PO), implying both the viability of these materials as coatings and 
their potential for targeting different GI regions. To allow for sampling in a more 
complex sequence such as that expected between the mouth and the small intestine 
(neutral ? acidic ? neutral pH, respectively), I layered 3 coatings each of L 100 and E 
PO sequentially over the capsule surface, depicted in Figure 3-10c. The resulting 
measurements with the sequence used, consisting of pH 7 to 3 to 7, are displayed in 
Figure 3-10d, where the increase in capacitance to saturation occurs at the return to 
neutral pH. This indicates that the E PO coating was able to protect the L 100 coating 
while transitioning to pH 3, while the L 100 coating ensured the gratings were sealed 
until the system returned to neutral pH. 
It is important to note that the slope between the baseline capacitance and the 
saturated capacitance varied with polymer type. This difference in timing was most 
likely affected by variations in each polymer?s dissolution rate. Alternatively, it is 
possible that it was also affected by their respective apparent viscosity at the solution 
concentration used (L 100: 60-120 mPa-s, S 100: 50-200 mPa-s, E PO: 3-6 mPa-s, 
according to the manufacturer), which varies the amount of plug infill during the initial 
coating and polymer remaining in subsequent steps. This would also impact the rate of 
solvent dissipation, specifically the methanol to air, for each coating. No observable 
80 
 
 
 
differences in coating appearance were seen during the coating process with each 
interval between coatings. 
81 
 
 
 
 
Figure 3-10 Sensor responses inserted into capsules with different types of coatings (3 dip-coatings each) 
compared to uncoated controls: (a) L 100, (b) E PO, and (c) combined coatings, where E PO is outer-most 
and L 100 is be-tween the 3-D-printed capsule and the E PO layers. Error bars depict standard deviation 
(n=4). (d) Capsule rendering for displaying coating conditions for each sequence. 
82 
 
 
 
3.3.3 Assembled Capsule Coating and Sensor Responses 
Assembled capsules were tested to ensure functionality of the fully integrated 
system. The insulated PCB with the sensor connected was inserted into neutral buffer, 
before insertion into the capsule, resulting in a spike and saturation in capacitance 
similar to the previous immersion experiments (data not shown). The current was 
monitored throughout to ensure the insulation of the board components was leak proof. 
The system was then inserted into the capsule and retested, producing the real-time 
capacitance readout (data not shown). The capacitance measurements produced from 
the polymer coated capsule in both neutral and acidic buffer are depicted in Figure 
3-11. Here, I observed that the sensors continued to measure the same capacitance from 
beginning to end for the neutral buffer, indicating that there was a lack of buffer inflow 
because the plug remained intact. Conversely, for the acidic buffer, a spike in 
capacitance is measured at about 23-27 minutes for each capsule, about 18-22 minutes 
after insertion into the solution. Interestingly, I observed that the change in capacitance 
is only ~20 pF compared to the 50 pF shift for the external electronics, which I found 
was due to the presence of additional parasitic capacitances introduced during the 
assembly process. However, there remains a distinct signal increase in the acidic pH 
compared to the neutral pH upon dissolution of the polymer coating. 
 For additional qualitative validation, the photographs in Figure 3-12 depict 
little to no change in the capsule coating throughout the neutral buffer sequence, 
whereas the coating is mostly dissolved ? both in the gratings the non-grating regions 
of the capsule surface ? as the experiment with the acidic buffer progressed. This result 
is consistent with the recorded capacitance measurements, indicating a specific time 
83 
 
 
 
point at which the buffer contacted the sensor. Based on the results in Section 3.3.1 and 
3.3.2, I could tailor this fluid intake for different timing windows by varying coating 
thicknesses, as well as coating materials.   
 
Figure 3-11 Real-time representative trials of capacitance measurements obtained via BLE from the 3x 
coated capsule immersed in control (neutral, pH 7) (red) and specific target (blue) pH (acidic, pH 3). The 
rapid increase in capacitance represents buffer inflow and contact with the sensors, due to coating 
dissolution. (n=2). 
 
Figure 3-12 Progression of polymer coating throughout (a) control and (b) pH-based dissolution tests. 
Beginning, 30 minutes, and 1 hour immersion periods are from the left, middle, and right, respectively. 
Scale bar = 3 mm. 
84 
 
 
 
 In this section, I described efforts to characterize the capsule coatings toward 
targeted fluid sampling in pH-specific regions. This was performed using our 
microelectronics system in a capsule platform, and was directed toward understanding 
the time scales achievable with different coatings thicknesses as well as different 
coating varieties through different pH sequences, i.e. from acidic to neutral, neutral to 
acidic, or a combination of neutral-acidic-neutral. This will be used toward sensing 
more specific molecules of interest, such as those present in the duodenum, while at 
target regions as described in chapter 4. 
3.4 Mechanical Testing 
 An additional feature considered for evaluating packaging efficacy was 
determining failure points for mechanical fracture. While the forces present during 
human peristalsis are relatively low, it is still important to perform some basic tests on 
the capsule to ensure it can withstand them [117], [118]. These forces vary significantly 
in the literature regardless of the GI region, ranging from 0.2 N in the duodenum to a 
high of 20.5 N during gastric emptying [198]?[200]. However, these forces can vary 
depending on the structure of the probe, and one work presents these forces through 
the scope of a magnetic pill, reporting significantly lower values than with this type of 
probe [201]. While the tensile strength of the material I used ? i.e. the MED610, 50-60 
MPa ? suggests it can withstand these forces, the structural integrity can vary based on 
features such as the layer thickness and number of layers used. Further, the capsule?s 
structure is inherently non-symmetrical, with the longitudinal length at 30 mm and axial 
length at 11 mm, thereby altering the contact surface area with the tissue and therefore 
the force distribution throughout the capsule body. 
85 
 
 
 
 To produce a basic evaluation of the force required to fracture the assembled 
capsule, a load frame was set up to apply force by compressing the capsule on either 
side from multiple axes. For each test, the 3D printed capsule bottom and top 
components were bonded together along the perimeter using epoxy (Devcon, Danvers, 
MA), then allowed to set over 24-hour periods to ensure complete curing. A diagram 
of the axes tested can be seen in Figure 3-13a below.  
 (a) (b)  
(c)  (d)  
Figure 3-13 (a) Diagram illustrating the regions along each axis where pressure was applied along the 
load frame. The colors are used for reference in (b) the applied force profiles for each setting. The failure 
point for each is indicated with each black circle. (c-d) Representative photographs of the capsule package 
upon reaching failure more for ?vertical? and ?on side? setups, respectively. (n=1 for each condition). 
 Figure 3-13b depicts each force profile generated with an Imada model MX 500 
load frame with a Z2H-440 2 kN load cell that has a load resolution of 0.1 kg 
86 
 
 
 
(Northbrook, IL), performing flexural tests with assistance from Mr. Nelson Quispe. 
As indicated with each circle, the failure points for each condition corresponding to the 
applied points in (a) were ~1189, 1681, 1040, and 301 N for the ?regular?, ?vertical?, 
?on side?, and ?upside down? positions, respectively, each of which far exceed the 
ranges anticipated during peristalsis. Figure 3-13c-d shows representative photographs 
of the failure modes for some of the setups, in which I observed failure points along 
the edges where the capsule surface had been epoxied. This was critical to understand 
asymmetries in the capsule assembly, which can ideally be improved through more 
robust and precise alignment procedures. 
 In this section, I determined potential failure modes for the capsule structure 
when stress is applied from different axes. While the failure force for each 
configuration exceeds expected values from peristalsis by, at the very least, one order 
of magnitude, the package rigidity could potentially be modified through more 
consistent applications of epoxy or other industrial sealants along the edges, and may 
even increase with the addition of the polymer coatings investigated earlier in this 
chapter. 
3.5 Conclusions 
In this chapter, I presented our investigation and characterization of packaging 
materials using a wireless microsystem capable of targeted fluid sampling based on 
pH-selective dissolution of polymer coatings. Due to the interface with the phone app 
and usage of off-the-shelf components for the circuit, this system enables a user-
friendly and cost-effective means of real-time data acquisition for GI fluid sampling 
dependent on targeted physiological characteristics. While further testing is critical to 
87 
 
 
 
understand the in vivo efficacy of the wireless electronics and signal acquisition, the 
system enables facile real-time evaluation of applied stimuli, which in this work 
consisted of dynamic pH profiles and their impact on various pH-targeting coatings. 
 Various limitations were noted during optimization and analysis. The 
conditions applied to test the coatings consisted of transient pH-sequences, whereas it 
is critical to test the response of each set of coatings, likely for those with thicknesses 
produced from 3-4 dip-coatings, over time at each distinct pH (3-7) to determine their 
respective abilities to maintain structural integrity. From a mechanical perspective, 
future capsule structures will likely be tailored by using fluid flow mechanical 
simulations to design enhanced inlet geometries that would enable more rapid sampling 
upon coating dissolution. The plug infill process could be improved to become more 
precise, accounting for viscosity of the polymer solution to improve uniformity across 
multiple coatings while continuing to maintain an effective seal. Furthermore, the 
slopes used to determine sampling rate may be affected by the presence of air bubbles 
in the sensing chamber, delaying the ability of the buffer solution to contact the sensor 
surface and thus reducing the accuracy of our assumed sampling rate.  
Our future work consists of using additional coating and plug materials that 
would react with more specific analytes present in GI fluids, such as peptide- or 
polysaccharide-based coatings which dissolve in response to digestive enzymes from 
pancreatic secretions. This platform will be adapted to have multiple sampling and 
sensing chambers, individually coated with a variety of biomarker-specific materials 
toward multiplexed and specific biomolecular identification and therefore enhanced 
utility and customization for diagnosing different GI pathologies. 
88 
 
 
 
4. Chapter 4:  Biochemical Capacitive Sensing with 
Triglyceride Films and Module Integration into Capsule 
System 
4.1 Capsule-Based System Integration 
In revisiting Table 1.1 and Table 3-1, various enzymes that play important roles 
to GI health are only present in specific regions, which can be targeted based on 
environmental pH. Devices designed to sense the enzyme in its target environment 
must therefore consider how best to isolate its signals relative to other potential stimuli. 
These stimuli can be primarily biochemical, such as the presence of other enzymes, 
ions, or different types of biomolecules, or physical, such as those from peristalsis or 
general body movement.  
 The capsule systems discussed previously utilize a variety of technologies for 
isolating their target signals to achieve specificity. The gas-sensing capsule, for 
example, utilizes gas-permeable membranes of graphene nanocomposites that are 
hydrophobic, aiding in phase separation of gas from liquid  [12], [13], [100], [202]. The 
rest of the capsule package, i.e. everywhere without the membrane, is made of a 
polyethylene shell that remains inert and protects the internal system from the GI 
environment. Meanwhile, one of the blood-detecting capsules is inherently specific due 
to genetically engineered E. coli that become luminescent in the presence of 
extracellular heme [102]. These devices as well as capsule endoscopes, however, do 
not have the challenge of performing functions at specific GI regions, as they are 
intended to function throughout the entirety of the GI tract. This often yields higher 
power requirements, which can lead to battery exhaustion if the operation time exceeds 
its limits, such as during capsule retention, therefore preventing procedural completion 
89 
 
 
 
[16]. When targeting markers that are only accessible in specific regions, the capsule 
does not need to function in irrelevant regions. Therefore, ensuring the device is only 
operating when certain environmental conditions are met, i.e. reducing the operation 
interval, can aid simultaneously in minimizing power consumption and nonspecific 
interference.  
Chapter 3 of this dissertation highlighted the portion of this work involves 
adapting a capacitive sensing system to know when the pH of the local fluid reached a 
certain level within the context of a capsule-package. Because the focus application of 
this dissertation is detecting pancreatic enzymes in the GI tract, system operation is 
confined to targeting pancreatic secretions after passing through the stomach, which 
possesses a pH of 1-2.5 and rises in the duodenum and small intestine to neutral levels 
according to Figure 3-2. In this chapter, I discuss our efforts to adapt our system to 
specifically sample secretions mimicking duodenal contents and subsequently measure 
pancreatic lipase and the impact of additional biomolecules that would be present. The 
system is low-cost, biocompatible, and easily interfaces with a mobile phone for 
wirelessly collecting data via BLE. The system offers promise for further development 
of ingestible diagnostic systems that would benefit from novel integration and 
packaging strategies for measuring biomarkers in GI secretions. 
4.2 Pancreatic Lipase Sensing Strategies 
 Pancreatic lipase (PL), as discussed throughout this work, is an essential 
hydrolytic enzyme for digestion and absorption of consumed triglycerides. Though 
other digestive lipases, such as gastric lipase (GL), also provide this capability, PL has 
a much higher specific activity ? 10,000 uM of free fatty acids (FFAs) compared to 
90 
 
 
 
1000 uM for PL and GL, respectively ? thereby contributing more potency and nutrient 
release [16]. Table 1-1 describes the various pathologies in which it becomes 
compromised, such as pancreatitis (both acute and chronic), adenocarcinoma, and 
general pancreatic insufficiency (umbrella label for reduction in pancreatic digestive 
enzyme activity), but may also occur in those suffering from cystic fibrosis or celiac 
disease [1], [2], [65], [82], [203]?[205].  
 Traditional means of PL evaluation are achieved most commonly using blood 
samples, though they can occasionally be tested in stool samples as well [206], [207]. 
Alternatively, PL can be measured directly from pancreatic secretions, collected 
through methods such as direct pancreas function testing (PFT) or endoscopic PFT. 
The prior utilizes an oroduodenal tube for access, then pancreatic stimulation via 
injection of hormones secretin or cholecystokinin (CCK). Unfortunately, this process 
is not standardized and the range is dependent on the center in which it is performed 
[208], [209]. Additionally, while exocrine insufficiency is more likely to become 
apparent during late stages of conditions like PA, it often contributes to malnutrition 
that can lead to a poorer prognosis even with surgery [210]. This implies, and has been 
shown, that treatments for exocrine insufficiency, i.e. pancreatic enzyme replacement 
therapy, can improve outcomes for PA patients [211]. Improving methods for 
monitoring exocrine insufficiency are therefore likely to aid in this process. 
Blood tests for lipase begin with centrifugation to extract the serum, then tested 
through a variety of methods such as colorimetric assay (570 nm wavelength absorption 
proportional to enzymatic activity) or titrimetric (pH of the solution indicates the extent 
of reaction). Pancreatic fluid samples, on the other hand, are only used to analyze 
91 
 
 
 
electrolyte concentrations due to accuracy limitations for enzyme measurements. These 
concentrations are determined using clinical chemistry autoanalyzers, which generally 
rely on ion-selective electrodes and measured voltage differences with varying 
concentrations of ions [209], [212], [213]. As a result, blood tests are chosen as the 
gold standards for lipase analysis, though are only useful when the condition severity 
requires stronger treatment. 
Novel methods for lipase detection have been emerging to improve upon 
current standards through features such as increasing accessibility and functioning with 
different sample types by removing required separation steps or hardware [214]. One 
sensor based on surface acoustic waves (SAW) monitored the frequency shift based on 
a change in solution conductance proportional to the lipase concentration. The solution 
conductance changed due to release of fatty acid from a triolein substrate [215]. 
Another sensor used screen-printed electrodes with immobilized Prussian Blue ? which 
reacts with H2O2 resulting from glycerol, a hydrolytic byproduct of triglycerides and 
lipase, and NAD+ after reacting with NADH oxidase ? for measuring the lipase 
concentration via amperometry [216]. However, these processes are relatively indirect 
and require either multiple reagents or highly specific conditions, likely limiting their 
ability to function as intended if immersed into GI secretions.  
Mirsky et al., however, describe an alternative scheme using short-chain and 
long-chain phospholipid substrates over gold capacitor electrodes to measure 
hydrolysis from phospholipase activity. The hydrolysis causes the substrate to become 
water soluble, thereby desorbing from the electrode surface into aqueous phase and 
increasing the electrode capacitance [217]. This strategy measures a more direct result 
92 
 
 
 
of the lipase activity, which may translate better to in situ environments. Furthermore, 
the strategy is readily adaptable to our microelectronics system developed in chapter 3 
of this thesis.  
However, the human duodenum is a host to not only pancreatic enzymes but 
incoming contents from the bile duct, specifically bile acids (BA). BAs are synthesized 
from cholesterol by hepatocytes; they are shown to be elevated in GI secretions during 
BA malabsorption and can be further related to conditions such as irritable bowel 
syndrome-associated diarrhea and colorectal cancer [218]?[220]. Because of their 
amphipathic nature, BAs such as cholic and deoxycholic acid act as natural detergents 
that emulsify triglycerides to form micelles through intercalation, aiding their digestion 
by PL into absorbable monoglycerides and fatty acids [221].Below, I describe our 
efforts to not only measure pancreatic lipase activity based on its ability to cause a 
triglyceride substrate to dissolve, but to characterize our system by testing our film with 
BAs to determine potential nonspecific impacts and how it may affect sensing of PL.  
 
4.3 System Setup 
4.3.1 Film Deposition 
 The sensing mechanism is based on the enzymatic and physicochemical 
reactions presented in Figure 4-1, which features (a) the hydrolysis of triglyceride ester 
linkages by PL and (b) the emulsification of triglyceride globules by BAs. Stearin was 
used as a model triglyceride due to its high melting temperature range (54-72.5?C) to 
remain stable at physiological temperature (37-39?C). The triglyceride used consist of 
a mixture of stearin and glycerol (SG), where several different ratios were tested. The 
93 
 
 
 
combinations tested include stearin, 2:1, 1:1, and 1:2 of stearin:glycerol after allowing 
to melt in a 100?C water bath prior to deposition. Multiple strategies for film deposition 
onto the sensors were tested, as depicted in Figure 4-2. The films used for testing with 
GI species were deposited via dip-coating of the sensor-PCB assemblies (SPAs, 
discussed in Section 3.2.3), depicted in Figure 4-2b. The SPAs were pre-heated for 5 
minutes to above substrate melting temperature (~100?C), then immersed and 
subsequently removed at ~10 mm/s. The films were air cooled at ambient forming SG-
coated SPAs (SG-SPAs), and resulting thicknesses were measured using calipers.   
The SG-SPAs were inserted into the capsule via the assembly process in Figure 
4-3, and the capsules were subsequently dip-coated into in pH-sensitive copolymer 
solution, depicted in Figure 4-2c. The copolymer solution utilized for this study was 
30% Eudragit L100 in methanol, producing coatings 783?60?m thick using a 
previously described coating process[195].  
(a) (b)  
Figure 4-1 (a) Lipase-induced hydrolytic digestion of stearin into glycerol and three stearic acid 
molecules. (b) Bile salt-induced emulsification of triglyceride globules into micelles. 
94 
 
 
 
(a)  
(b)  
(c)  
Figure 4-2 Deposition strategies for obtaining TG films over sensors: (a) drop-casting and (b) dip-coating 
of molten TG solution while the substrates were either left at ambient or pre-heated to TG melting 
temperature for 5 minutes. (c) Dip-coating of assembled capsules into dyed pH-sensitive copolymer 
solutions to form GI-targeting coatings, repeated for sequential coatings. 
95 
 
 
 
 
Figure 4-3 Assembly process for platform. Steps 1-4 used for sensor characterization of TG-SPAs with 
connection of VDD and GND to external power supply, while Steps 5-8 are used for testing complete 
capsule. At Step 5, after connection to Lithium polymer battery. Deep Sleep mode is entered for a 3-hour 
time period via transmitted signal to allow for curing of adhesion and coating materials at respective 
steps. After Step 8, capsule electronics enter Active mode to enable capacitance measurements and signal 
transmission to mobile phone. 
 
4.3.2 Sensing Characterization 
SG-SPA characterization was performed while powered by an Agilent E3631A 
DC power supply (Santa Clara, CA).  For the SG-SPAs prepared in Section 4.3.1, each 
experiment was initiated with baseline measurements while the device was suspended 
in air. After 5 minutes, the devices were lowered into a 50 mL beaker containing 
solutions of either negative control ? i.e. buffer only ? then subsequent test conditions, 
described below. All buffer solutions consisted of 40 mL of 0.1 M phosphate buffered 
saline (PBS) on a hot plate set to 300 RPM stir via magnetic stir bar to equilibrate the 
solution and temperature at 39?C. All solutions were prepared using deionized water 
96 
 
 
 
from E-pure Ultrapure Water Purification Systems (DIH2O; resistivity=18.0 ?-cm; 
Thermo Scientific, Waltham, MA). 
The sensor capacitance response was determined for each analyte in an 
environment that simulated fluids present in the duodenum. SG-SPAs were immersed 
in pH 7.3 buffer containing varying concentrations of either porcine PL or a BA mixture 
of sodium cholate and sodium deoxycholate (Sigma Aldrich, St. Louis, MO). PL was 
tested at 1 mM, 100 ?M, and 10 ?M concentrations, while BA was tested across 7, 0.7, 
and 0.07 %w/v. To measure potential impact of nonspecific enzymes, porcine 
pancreatic trypsin (Sigma) was also tested alone at 100 ?M, a concentration that 
exceeds maximum expected outputs by >250 % [222]. Each solution was incubated for 
30 minutes at 39?C prior to insertion into a beaker for testing the SG-SPAs to ensure 
complete solubility. 
4.3.3 SEM Analysis for Film Morphology 
Pre-heated sensor die were inserted into molten 2:1 SG solution prepared as 
described in Section 4.3.1. The sensors were then cooled at ambient (23?C) for 24 
hours, then incubated in glass petri dishes containing PBS (0.1 M) alone or with either 
100 ?M PL, 0.7 %w/v BA or 100 ?M trypsin, respectively. Each solution was 
maintained at 39?C under 300 rpm stir. Sensors were removed from solution after either 
30- or 60-minutes, rinsed with DIH2O, dried for 24 hours, then prepared in a carbon 
coater to deposit coatings of conductive carbon (MED 010 Balzers Union Carbon 
Coater, Balzers Union, Liechtenstein). The samples were then viewed under a Hitachi 
S-3400 scanning electron microscope (SEM). 
97 
 
 
 
4.3.4 pH-Dependent Sampling and Sensing 
Following a similar procedure as in Section 4.3.2, measurements were initiated 
while suspending the coated capsule in air and powered by a Li-polymer battery 
(Powerstream, West Orem, Utah). The capsules were then sequentially inserted for a 
duration of 25 min into several solutions, which mimicked the pH transition between 
the stomach (acidic) and duodenum (neutral), as depicted in Figure 4-4. The solutions 
consisted of the following conditions: (i) 0.1 M acetic acid (pH 3), (ii) PBS (pH 7.3), 
and (iii) 1 mM PL in PBS (pH 7.3). This sequence was chosen to reflect the capsule 
transit throughout the GI tract with the subsequent presence of a specific biomarker 
residing in the SI, in this case PL. 
(a)  (b)  
Figure 4-4 (a) Schematic depicting normal GI pH progression along with adjacent organs. (b) 
Experimental setup for complete capsule characterization. 
4.4 Results and Discussion 
This work presents a sensing strategy using triglyceride films to detect duodenal 
contents, such as lipase and bile acids, in a wireless capacitive-sensing platform. The 
sensors were designed to measure the capacitive response when fluid enters the sensing 
chamber. Under biological stimuli from species in the environment, the substrate films 
deposited over the sensors would undergo reactions, such as hydrolysis or 
98 
 
 
 
emulsification, that alter their dielectric property. Gradually, the dissolution of the film 
substrate exposes the electrode fingers to the infiltrating fluid, causing an increase in 
capacitance. The sensor capacitance was measured for SG-SPAs and coated capsules 
inserted into various solutions containing biochemical stimuli that would induce 
changes in capacitance over different lengths of time when compared to non-specific 
controls or buffer alone.  
Stearin was the material chosen as our model triglyceride, as described in B. 
Triglycerides, in general, are significant to traditional human diets, and its digestion is 
enabled due to emulsion and subsequent hydrolysis by BAs and PL, respectively. Due 
to its rigidity and low water solubility ? an effect of having longer chain fatty acids ? 
unmodified stearin is an inefficient substrate for hydrolysis. Other triglycerides such as 
oleic acid ? a shorter chain fatty acid found in olive oil ? are used as standard substrate 
for lipase assays; however, these substrates are incompatible with our system function 
in the small intestine due to having a melting point below physiological temperature 
[223]. Previous reports modify stearin as a substrate using suspensions with glycerol 
by enhancing the quality of the interface it would have with lipase [224]?[226]. Lipase 
activity on triglycerides is generally dependent on the surface area, which increases 
significantly with rougher surfaces or when emulsified into micelles by BAs. While 
crystalline stearin films are naturally hydrophobic, glycerol is hydrophilic due to polar 
?OH groups, enhancing the interface between the species in solution and the film 
substrate, as well as a stable film deposition and adhesion over the sensor surfaces. 
99 
 
 
 
 Because the platform has demonstrated utility for monitoring changes in sensor 
capacitance (C), I could characterize dielectric properties of the films through 
estimation by Equation (9): 
2???0??? ???0? (9) 
? = =  
? ?
This is similar to equation (7) back in Chapter 2, where it uses ?0 (vacuum 
permittivity), ?r (dielectric constant), l (finger length), w (finger width), n (numbers of 
electrode fingers), and d (distance of separation), where 2lwn can be simplified to A 
(total area of both electrodes). Therefore, changes in the dielectric properties of the 
films are proportional to measured capacitance (C), and indicate the extent of 
hydrolysis or emulsification of the substrate over the sensor. To account for changes in 
baseline capacitance between experiments, I presented sensor response as the percent 
change in capacitance (%?C), which is calculated with equation (10): 
?? ? ?1
%?? =  ( ) ? 100 (10) 
?1
When no film is present ? i.e. one of our negative controls ? and the sensor is 
inserted into PBS, the mean %?C is 39.5?1.4, likely a resulting effect of increase in 
dielectric constant combined with ionic reactions with the SiO2 films, which can be 
mitigated in the future through more inert films [227]. As expected, insertion into 
molten and cooled SG solutions produced negligible %?C, indicating that changes in 
capacitance are limited to (1) when the SG material is influenced by an environmental 
condition ? such as the applied solutions and constituent analytes described in C ? and 
(2) due to downstream effects on the SiO2 films once the insulating integrity of the SG 
films becomes compromised, though this latter condition is not studied further here. 
100 
 
 
 
I first tested the effect of different deposition strategies of the SG solutions onto 
our sensors, depicted in Figure 4-5. Drop-casting and dip-coating of molten SG 
solution, while the substrates were either left at ambient or pre-heated for 5 minutes to 
SG melting temperature, were compared. Substrate temperature was considered a 
significant factor due to its effect on surface tension, and therefore wetting of the 
solution [228]. In the case of drop-casting, three phases (i.e. solid, liquid, gas) are 
present at the sensor surface during the entirety of the process, thus surface tension 
becomes a more dominant factor for wetting, and therefore coating, the sensors. 
Wetting through dip-coating, however, causes most of the sensor surface (except for 
the edges) to be exposed to only two phases with the ability for a complete wetting and 
sealing layer, reducing the dependency on surface tension. Additionally, the order of 
the wetting transition has a direct impact on the adsorbed film thickness, such that the 
greater the discontinuity in the interfacial energy between phases, the greater the 
thickness. This discontinuity is reflected by the difference in temperature between the 
substrate and film, which I observed through thinner films produced when the 
substrates are pre-heated. Additionally, the SiO2 films on the sensor surfaces are 
generally hydrophilic, with surface energies of ~73.8 mJ/m2, compared to most 
triglycerides, which maintain surface energies ranging from 25-30 mJ/m2; this 
increases with temperature as well, producing further discontinuity and, therefore, 
wetting angle [229], [230]. Adhesion between the SiO2 and SG layers is primarily based 
on van der Waals interactions; however, I found that the drop-casted films were not 
stable, and failed to prevent buffer from interacting with the sensors (data not shown) 
[231]. After comparing the efficacy of the film deposition methods, SG-SPAs were 
101 
 
 
 
tested using films produced through pre-heating the substrate to 100?C and dip-coating, 
which yielded an average film thickness of 210?60.3 ?m. The total sensor area was 
0.06 mm2, indicating each film volume over the sensors to be 1.25x10-5 cm3. In 
comparing the sensor response for different SG film compositions (see C), I found the 
2:1 SG ratio films to be most stable when in buffer alone (Figure 4-6), hence their 
implementation in the following sensor characterization experiments. From the 2:1 SG 
ratio, I could calculate the volume of stearin contained within the film to be 0.84x10-5 
cm3, and with a density of 0.862 g/cm3, the stearin mass was found to be 0.724x10-5 g, 
or  8.12x10-9 mol (molecular weight of stearin:891.5 g/mol). 
 
Figure 4-5 Resulting thicknesses in films with each deposition strategy and initial SPA temperature. Error 
bars depict standard deviation (n=3). 
102 
 
 
 
 
Figure 4-6 Representative sensor response testing after insertion of TG-SPAs into buffer solution (0.1 M 
PBS) when TG film composed of either pure stearin or stearin:glycerol (S:G) ratios of 2:1, 1:1, or 1:2. Film 
most stable, i.e. least response over time, with the 2:1 S:G ratio compared to other compositions. (n=2). 
The standard concentration of PL expected in healthy unstimulated duodenal 
fluid ranges from 1-10 ?M [232]. Conversely, average duodenal secretions contain 0.7 
%w/v BA, which reflects the median concentration tested in this work [233]. Because 
PL is dependent on BAs for improving the enzyme-substrate interface with 
triglycerides, it was necessary to determine the individual effect of both species on the 
dielectric properties of the film, which can range from characteristics such as ratio of 
stearic acid to glycerol via hydrolysis, stearin to glycerol through emulsion separation, 
or even overall SG film to solution contact with the sensor surface. Figure 4-7 presents 
the capacitive response (%?C) of the SG-SPA over time upon immediate insertion into 
the solution at t=0 minutes, where the concentration of the biological stimuli was varied 
for PL (0.01-1 mM) and BAs (0.07-7 %w/v), as well as pancreatic trypsin (100 ?M) to 
measure potential nonspecific interactions. However, it must be noted that these are 
representative trials and more runs are required to determine film robustness and 
reliability to different concentrations of lipase. Each sequence begins directly following 
103 
 
 
 
insertion of the SG-SPA into the test solution. Measurements were compared to a 
buffer-only solution (7.3 pH, PBS) as a negative control. The capacitance responses of 
films to both PL and BAs produced trends that altered proportionally to the 
concentration of the analytes present. The measurement was conducted over the course 
of 40 minutes, consistent with the expected transit time of most contents passing 
through the duodenum, though longer time scales have been reported [234]. 
Additionally, nonspecific activity testing with extracted duodenal secretions and its 
regular contents will be necessary for increasing confidence in sensor specificity, 
though stearin appears to be a resilient insulator to negative controls. To modify the 
system for response to lower concentrations, it is likely that parameters such as 
concentration of substrate at the sensor surface or film thickness will need to be reduced 
such that the dielectric properties of the film can change faster. This is consistent with 
our findings where films exceeding 200 ?m thickness produced no significant change 
in capacitance compared to buffer alone (data not shown), indicating that increasing 
film thickness may reduce reactivity or penetration of the enzyme. Reducing the film 
thickness would be feasible through closer matching of the surface energies, such as 
through temperature or surface tension, between the SPAs and SG solution before dip-
coating, as well as dissolving the triglyceride in nonpolar solvents such as ethers, 
hexane, or chloroform in lower concentrations [78], [235] 
 
 
 
 
104 
 
 
 
(a)  
(b)  
(c)  
Figure 4-7 Sensing in the presence of different biochemical species at varying concentrations. (a) PL, (b) 
BAs, and (c) pancreatic trypsin. 
After measuring the capacitive response to each concentration, I used the slopes 
of the capacitive response for the first 30 minutes of sampling (%?C/?T) as a metric 
105 
 
 
 
to fit the sensitivity. While we cannot model the data without more trial runs in future 
works, this provides insight as to the kinetic efficacy of each analyte to the substrate, 
which may change due to the difference in reactive mechanism for each analyte, i.e. 
hydrolysis and emulsification for PL and BAs, respectively. For example, hydrolysis 
via PL produces glycerol and stearic acid, each of which impact the capacitance of the 
sensor interface at both the SG and SiO2 films differently. Alternatively, emulsification 
does not alter the material chemical composition, merely the particle size and therefore 
phase properties of the film, allowing enhanced exposure of the SiO2 films and 
subsequently electrode surface to constituents in the environmental fluid, even those 
such as divalent cations [236]. These results produce insight as to the existence of 
potential differences in the dielectric impact between hydrolytic and emulsion kinetics, 
though further investigation is required to yield specific electrochemical effects of GI 
analytes on these individual materials. 
Morphological differences comparing the impact of each species on the SG 
films were viewed under SEM (110x magnification), after deposition of a carbon 
coating for conductivity. The films had been incubated under stir condition to each 
solution, and removed at two time points, 30 minutes and 60 minutes, as shown in 
Figure 4-8. The 30-minute time point was used to correspond the film morphology to 
the endpoints of the %?C/?T values used earlier, while the 60-minute time point was 
used to determine if any significant changes in the film morphology occurred beyond 
the used duration. Here, I observed little difference between the samples exposed to 
either PBS or trypsin, whereas the surfaces for PL and BA indicate significant 
qualitative changes in roughness and crystallization, respectively. For the PL-exposed 
106 
 
 
 
sample, the surface appears significantly smoother, likely an effect of the hydrolytic 
interface occurring between PL and the stearin that reduces surface roughness, and 
therefore interfacial surface area, until eventual saturation. The BA-exposed sample, 
however, presents the formation of an increasingly fragmented stearin surface, offering 
insight as to how hydrolysis and emulsification manifest differently on the substrate. 
Based on the capacitive changes observed in Figure 4-7, these structural changes may 
be direct indicators for changes in either dielectric properties of the film material or, 
through emulsification-induced fragmentation, surface area of the underlying 
electrodes to the high-dielectric behavior of the buffer. The lack of significant change 
between the 30- and 60-minute time points also imply that the durations used are 
enough to correspond signal saturation to the morphological state. 
107 
 
 
 
 
Figure 4-8 Representative SEM images for characterization of SG films after exposure to various solutions 
at 30- and 60-minute time intervals. (n=3). 
108 
 
 
 
Finally, I tested the ability to measure PL within the capsule to simulate 
detection in the duodenum environment. Figure 4-9 presents the result of testing the 
coated capsule containing the SG-SPAs for performing pH-targeted sampling. The 
respective conditions are labeled as acid, neutral, and lipase, where the capsule was 
inserted into a pH 3 solution (0.1 M acetic acid) to represent acidic contents for the 
stomach, pH 7.3 solution (0.1 M PBS) for the duodenum, and addition of PL (1 mM 
PL in 0.1 M PBS) as a target analyte, respectively. Here, the sensor responses are 
overlaid for more direct sequence comparison, and presented in capacitance rather than 
%?C to show how they compare from beginning to end of each condition. Upon 
insertion into the acid condition, I observed negative drift from ~129.5?0.772 pF, 
eventually saturating at 115.2?0.759 pF, though no inflow of solution into the capsule 
was evident through visual inspection and the color of the buffer solution remained 
clear as indicated in Figure 4-9b-I [195]. After ~19 minutes of incubation in the neutral 
condition (0.1 M PBS), a distinct increase in capacitance to 146.1 pF was observed, 
before reaching saturation at 127.9?0.433 pF. Additionally, the color of the solution 
gained a red hue reflective of the Eudragit L100 coating color, as seen in Figure 4-9b-
ii, indicating sample entry. Once the capsule was inserted into the lipase condition, I 
observed an immediate effect similar to the capacitance sequences observed in Figure 
4-7a, where there is a rapid increase in capacitance over ~14 minutes to reach 
216.0?0.602 pF, or 66.8% increase from the beginning, with some signal fluctuation 
and decrease as well until it levels around 158.5?0.913 pF, or a 22.4% increase. Due 
to the opacity of the PL solution, I did not discern any distinguishable features until 
109 
 
 
 
capsule removal, where there is a noticeable region in the Eudragit L100 coating where 
the fluid could enter the capsule sensing chamber, shown in Figure 4-9b-iv. 
 (a)  
(b)  
Figure 4-9 (a) Combined pH sampling and enzyme sensing in integrated capsule. Measurements were 
initiated after immersion into acidic solution for 25 minutes (blue), after which the capsules were 
removed and immersed into a neutral solution (red), where a spike was observed around 19 minutes, 
indicated by the circle. Finally, the capsule was removed and immersed into a solution containing 1 mM 
lipase (black), where a distinct increase in capacitance is observed, indicated by the arrow. (b) 
Photographs taken of the capsule at the end of each respective sequence, designated by the label color: b-
i: acid condition, b-ii: neutral condition, b-iii: lipase condition, and b-iv lipase condition after removal 
from the solution. (n=1). 
110 
 
 
 
I then calculated the slope from the first 30 minutes of the lipase condition to 
yield 0.198 pF/min. Comparing the linear fits of the slopes show changes with 
concentration of PL. This slope is expected when the concentration is ~160 ?M, even 
though the applied concentration was 1 mM. This is calculated as a 6.25-fold loss in 
sensor response, indicating loss in SG-SPA response to the analyte when packaged 
within the capsule rather than when tested without it; this is not surprising, though, 
considering the differences in fluid dynamics and flow profile of the analyte in the 
sensing chamber as opposed to when the SG-SPA is surrounded by the fluid in the 
beaker. However, this emphasizes the need to enhance the exposure of the SG-coated 
sensors to the surrounding environment, which can be implemented by reducing the 
thickness of the 3D-printed shell at the inlets or by embedding the sensors directly onto 
the outermost packaging of the capsule. Ultimately, this experiment demonstrated the 
sequential ability of the capsule package to remain intact in a potential gastric 
environment, dissolve from duodenal fluids and allow detection of PL in the same 
environment. Based on the results presented in Figure 4-7, it is likely this similar 
strategy can be leveraged to detect BAs in target pH environments as well, as well as 
for measuring PL or BA related pathologies such as those discussed above. 
 In this section, I described our efforts to develop various triglyceride film 
deposition protocols, including the impact of composition and method on stability, and 
the subsequent integration into the capsule system and pH-soluble coatings discussed 
in chapter 3. Further, we gained an understanding of how the sensing response can be 
applied to both hydrolytic and emulsification reactions, which possess utility for 
detection of a broader range of targets than initially intended. While further efforts 
111 
 
 
 
should be established in enhance specificity to individual reactions at a time, this 
maintains significant advancements to the potential impact of capsule-embedded 
sensing technologies. 
4.5 Conclusion 
This work presents a proof-of-concept demonstration of triglyceride-based 
coatings for monitoring duodenal analytes in an integrated capsule system. The system 
can measure and wirelessly transmit changes in capacitance due to both film hydrolysis 
and emulsification reactions as a result of exposure to PL and BAs, respectively. For 
PL, I measured signals for 10 ?M-1 mM concentrations, whereas for BAs I could 
distinguish between 0.07, 0.7, and 7 w/v% concentrations, each of which are within 
physiologically relevant ranges for healthy and abnormal levels. For each 
concentration, I used the observed slope produced from the first 30 minutes, which is 
well within the average duration range for material traveling through the human 
duodenum, to relate the signal to its respective concentration [234]. Furthermore, I 
characterized morphological differences in triglyceride-based films after exposure to 
various duodenal contents, giving insight as to the physical manifestation of hydrolytic 
versus emulsification reaction mechanism. Finally, I demonstrated the ability to detect 
PL after subsequent targeting at a pH-specific environment, and found that the design 
of the sensing chamber and the respective inlets could be thinned to prevent reduction 
in the reactive interface between the sensor surface and environmental contents. 
There are several limitations to our methods for signal analysis and the platform 
in general. For example, parametric analyses of slopes over different time intervals may 
allow for distinguishing between each film response over time, thus producing more 
112 
 
 
 
sensitive or optimal fitting equations to match the concentration with the %?C/?T. The 
sensor responses are also likely to increase sensitivity if the insulating films can be 
deposited with reduced thicknesses; this would change the amount of substrate and 
therefore amount of reaction that needs to occur to increase the rate of change in 
dielectric mismatch between the film and environmental solution, potential altering 
resulting concentration range for other GI regions where it may be relevant. Next,  each 
species was tested in isolation, and their simultaneous presence has been shown to 
reduce efficacy due to non-specific film interactions depending on the material 
used[144]. Therefore, it is critical that future work will focus on characterizing and 
optimizing the sensor performance over a wider concentration range, adding more non-
specific species to determine further reaction specificity, and incorporate additional 
food-based materials, such as gelatin ? substrates for monitoring other types of 
pancreatic enzymes [144]. Additionally, different capsule coating materials could be 
used toward measuring these species in different GI environments ? based on their pH 
? as well as enhance the exposure of the substrate to materials in the external 
environment. Finally, the sensor response can be utilized to trigger the release of on-
board actuation mechanisms or drug payloads for feedback-enhanced control over 
targeted therapies. Ultimately, this platform provides opportunities for sensing 
hydrolytic and emulsification reactions, such as through other enzymes or biologicals 
similar to the species discussed in this work, along with an innovative packaging 
strategy toward passive GI targeting in minimally invasive and ingestible diagnostics.   
113 
 
 
 
5. Chapter 5:  Concluding Remarks 
Highlights 
? Electrical characterization of film-based sensing strategy for bioelectronic 
interfacing 
o Distinguished impedimetric and geometric trends produced from 
specific and nonspecific film interactions  
o Corresponded impedance trends with enzyme concentration 
o Tailorable for a variety of films for different reactant-substrate 
interactions 
o Characterized for different sensor geometries 
o Determined potential signal metrics for integrated sensing systems 
? Development of coating-based packaging strategy to enable gastrointestinal 
targeting of microsystem platforms 
o Targeting demonstrated for environments with different pH range 
o Suggestive for targeting gastrointestinal regions such as the duodenum 
and stomach 
o Adaptable for multi-layered coatings toward dissolution after 
sequential pH ranges 
o Demonstrated resilience to expected gastrointestinal mechanical forces 
? Demonstration of microelectronics sensing system for detecting duodenal 
targets with triglyceride films 
o Various deposition strategies for triglyceride materials 
114 
 
 
 
? Evaluated with dependency on sensor substrate and material 
composition 
o Characterized morphological changes occurring at the microscale in 
response to hydrolytic vs emulsification reactions 
? Sensor response similarities between pancreatic lipase and bile 
acids 
o Reduced response for nonspecific enzymes 
? Integration of sensing system within capsule package demonstrates feasibility 
for pH-specific environmental targeting and subsequent sensing 
o Distinguished between sampling signal and specific marker signal 
o Developed assembly and validation protocols for device function 
5.1 Summary 
This study presents the design and characterization of various critical elements, 
summarized in Table 5-1, required in integrating sensing and targeting strategies into a 
capsule platform for evaluating GI health. This work explored a capacitive-based 
sensing and detection method relying on the reaction of deposited biomaterial films 
with specific biomolecules, specifically pancreatic trypsin, lipase, and bile acids, within 
the context of a capsule-sized microelectronics system capable of wireless, real-time 
monitoring. The first experimental chapter, i.e. chapter 2, described the development 
of a proof-of-concept method that measured impedance of these films and isolated 
specific relevant electrical elements that could be used to design a microelectronics 
system with capsule-sized form factor, and investigated various deposition strategies 
for integration with traditional microelectromechanical system (MEMS)-based 
115 
 
 
 
fabrication processes. Chapter 3 then presents our passive approach to targeting 
specific GI fluids based on their pH profile. This was achieved using coatings ? 
consisting of copolymers utilized in the pharmaceutical industry for targeted drug 
delivery functions ? that were repurposed for protecting our sensor surfaces in 
nonspecific pH environments and dissolving, thereby exposing the surfaces, at the 
regions of interest. Specifically, I tested the ability to sample in acidic (pH < 3) and 
neutral/basic (pH ? 7) environments, and used a novel microelectronics capacitive 
sensing platform, coupled with wireless capabilities for interfacing with a mobile app, 
to characterize the sensor response when the coatings were implemented over 3D-
printed capsule structures used for housing the sensors. Finally, in chapter 4, I 
demonstrated the platform for measuring triglyceride films and their responses to 
various duodenal contents such as pancreatic lipase and bile acids, as well as its 
integration within the coated capsule package described in chapter 3. Figure 5-1 
illustrates the prospective data flow from the capsule device, where wireless data 
extraction enables external access to the status of the potential concentration of the 
biomolecule of interest at the target region to the patient?s physician, offering insight 
as to critical information about their health that would aid prognosis. 
 
 
 
 
 
 
116 
 
 
 
Table 5-1 Summary of system parameters for the capsule device with reference to their respective 
chapters for more detail. 
Module Characteristic Metrics Chapter 
Packaging (Outer) Eudragit (Dip-coating) 35 mm x 12.7 mm 3  
MED610 (3D-Printing) 
Packaging (Inner) Conformal Coating 33.8 mm x 11.5 mm 3,4 
Silicone Sealant 
Epoxy 
Sensing Method Capacitance 0.8-220 pF 3,4 
7.3 pF/mV 
Sensing Trypsin (Sensor Only) Gelatin/Glutaraldehyde 2,4 
Targets/Substrates Lipase/Bile Acids Stearin/Glycerol 
(SPA/Capsule) 
Wireless Comm. Bluetooth LE 2.4 GHz 3,4 
Power Source Li-Polymer 14 mAh/3.3V 3,4 
 
 
Figure 5-1 Depiction of application. The capsule is protection in environmental pH reflecting that of the 
stomach until reaching the small intestine via pH-sensitive polymer coating. Upon dissolution, the fluid 
is sampled via embedded gratings, and pancreatic lipase reacts with triglyceride coatings to increase the 
capacitance of the sensors. This is transmitted to a phone wirelessly, and the data can be shared with 
medical practitioners. 
117 
 
 
 
In chapter 2, I first studied gelatin films, specifically methods for their 
deposition and crosslinking for physiological stability and their overall integration with 
microfabricated impedance sensors for responding to various pancreatic enzymes. 
Using IDE sensor designs initially intended for measuring biofilm growth, I was 
interested in observing their utility for measuring biomaterial film processes that 
involved their reduction instead [142]. After preliminary results involving impedance 
trends ? such as those over spectra from 10 Hz-1 MHz at specific frequencies and 
through Nyquist analyses ? specific to our target enzyme, i.e. pancreatic trypsin, I 
compared them to trends produced when either the film was absent and/or in the 
presence of nonspecific enzymes (lipase and amylase). This yielded observable 
distinctions, such as those seen in Figure 5-2, between them that I then used to optimize 
sensor design parameters, which involved characterization of various process 
parameters and resulting films that could be produced with different spin speeds and 
device geometries while determining feasible sensor parameters that could enhance 
sensitivity. Impedance results were also decomposed into real and imaginary elements, 
enabling our transition to measuring capacitance with a subsequently developed 
capacitive-sensing and wireless electronics platform. 
 
Figure 5-2 (Left) Schematic and (right) optical images of sensors with deposited film. The optical images 
indicate distinct removal of film after trypsin response compared to buffer alone, where the film remains 
relatively intact. 
118 
 
 
 
Chapter 3 consists of using the platform, after thorough assembly and 
calibration processes, to evaluate a novel packaging strategy with a capsule form factor 
to enable passive targeting of the capsule into simulated target GI regions based on 
their pH. Various capsule geometries were designed and considered for testing, each 
with some feature that overcame one of the prior iteration?s flaws; each design required 
at least inlets that allow inflow of environmental fluid into a chamber containing the 
sensors. Eudragit E PO, S100, and L100 polymer coatings were tested for targeting 
simulated fluids for the stomach, duodenum, and ileum, respectively, with an emphasis 
on coating thickness and measuring the time of dissolution and subsequent filling of 
the sensing chamber. The platform was transitioned into a microelectronics system, 
depicted in Figure 5-3, with a small form factor for subsequent integration into a 
capsule system, after which additional assembly testing led to its validation as an 
independent system, powered with a Li-polymer battery and controlled through 
commands from a mobile app. As described in chapter 3, the primary utility of the 
system is not just through its potential for in vivo use, but for enabling user-friendly 
and real-time measurement acquisition through different applied stimuli and their 
impacts on various system characteristics. 
(a) (b)  (c)  
Figure 5-3 (a) Top and (b) bottom views of sensor-connected PCB, and (c) photograph of assembled 
capsule. Scale bar=5 mm. 
Based on the fundamental approaches we developed using capacitance as a 
metric for measuring biomaterial film reactions with specific and nonspecific targets, I 
119 
 
 
 
further adapted the sensors used in the microelectronics system, or SPA, to measure 
triglyceride film reactions in chapter 4. Stearin films were deposited in molten phase 
and with glycerol as a solution modifier and subsequent cooling using multiple 
approaches, with dip-coating producing the most stable to the implemented 
physiologically simulating environments. The amorphous films were characterized, 
offering signal profiles that also suggested dependencies on film qualities, and 
observed to affect electrical properties of the sensor surface by measuring capacitance 
and morphology, with the SG-SPAs and via SEM, respectively. These features were 
analyzed in response to hydrolytic reactions from pancreatic lipase as well as 
emulsification from bile acids. The capacitance response over time was found to vary 
with concentration, and while this implies that neither signal can be isolated if both are 
simultaneously present, this also reduces limitations as to the types of reactions that 
can be measured with the platform. Furthermore, the SG-SPAs were then assembled 
within the Eudragit coated capsules, which were observed to protect the sensors in 
acidic conditions until the environment reached neutral conditions. Upon sampling, 
application of PL into the solution produced more significant changes in capacitance, 
reflecting the ability to perform pH-targeted sensing with the encapsulated SG-SPAs.   
5.2 Future Outlook 
 The work presented in this dissertation has led to the emergence of new 
directions that can be pursued for either enhancing characterization methods for capsule 
devices or additional features for more complex functionalities. 
120 
 
 
 
5.2.1 Gastrointestinal-Simulating Biochemical and Biomechanical Conditions  
 Predicting capsule function in vivo through benchtop experimentation relies 
significantly on how well the testing platform mimics the target environment. The 
capsule geometry has been most utilized because it is already implemented in 
pharmaceuticals for daily use, but can be implied that it limits disruption to most GI 
functions. Most benchtop tests, such as those performed in this dissertation, consist of 
beaker-level testing and isolated or low complexity solution compositions until they 
are implemented in animal models for the next phase of device characterizations, 
specifically porcine due to its distinct fundamental similarities to humans in terms of 
anatomy, physiology, and nutrition, among others [237]. However, more complex 
functions desired in later capsule designs may produce more risk to the GI environment 
of the animal, while there also may not be enough comparison between them and 
humans to imply downstream clinical success. This leads to costly experimentation 
requirements that can be remediated through lower-cost testing setups to provide more 
relevant results and expectations. 
 A primary feature of the GI tract that can alter device function is the state of its 
biochemical composition over time, especially considering its variability upon 
stimulation through either intravenous administration or oral ingestion. A few 
examples, as discussed in chapter 1 of this thesis, of how the output can change is 
through tests such as the secretin-CCK and Lundh tests, where injected hormones or a 
carefully devised meal can stimulate increased pancreatic secretion output for higher 
sampling rates and enzyme concentrations in the duodenum [55], [126]. While such 
simulants only account for changes in pancreatic output, it is critical to evaluating 
121 
 
 
 
device function that the device be tested in the presence of other GI secretions, such as 
those present in the gastric environment. To compensate for this variability, some 
procedures implement methods of establishing control over GI content, which can be 
observed in the preparation for capsule or normal endoscopic procedures. Here, the 
patient is instructed to fast overnight, cleanse with polyethylene glycol (PEG) solutions, 
or drinking clear solutions such as water to ensure visualization remains unimpeded 
[238].  
For the device presented in this dissertation, it is likely that future developments 
will require the establishment of similar preparations while assessing how different 
stimulants can affect device efficacy. This efficacy, however, must be evaluated from 
multiple perspectives. Most relevant to the characterizations performed in this work, 
however, are the impact of such preparations on (a) sensing and (b) packaging or 
targeting efficacies. For in vitro characterizations, fluids should be altered periodically 
through various conditions ? such as aliquots of enzymes presented in Table 1-1, 
various local PEG concentrations, or different pH levels for example ? and how the 
system responds. For (a), it must be understood how film-coated sensors react both 
through resulting capacitive signals, as well as potential morphological changes, 
including the impact on film mass and thickness. For (b), sensors without films must 
be tested with these fluids, when integrated into capsules coated with the Eudragit 
formulations described in Chapter 3, to determine how the timing of coating dissolution 
can behave. Ultimately, both sensing and GI-targeting functions should be tested when 
film-coated sensors are integrated within coated capsules, and capacitance signals 
should be monitored when each stimulus is applied in real-time.   
122 
 
 
 
 From a mechanical perspective, models for replicating the forces ? such as 
traction, contact, and peristaltic wave propagation ? produced in GI peristalsis have 
been designed for capsule devices [239]. Much work has been done on understanding 
the mechanics of various regions of GI propulsion, with robotic capsule devices aiding 
in this evaluation by providing metrics for their own motility [240], [241]. For this 
dissertation, simulating intestinal propulsion would aid significantly in evaluating the 
integrity of the capsule coatings, enabling us to implement further design changes that 
could affect their respective efficacies, especially in both specific and nonspecific pH 
environments.  
Figure 5-4 presents a concept developed by Prateek Sayyarapaju and I, 
consisting of 3D-printed components for intestinal peristalsis simulation. The advent 
of 3D-printing has enabled facile fabrication of numerous components types, varying 
in characteristics such as rigidity, opacity, and biocompatibility, such that we can vary 
the material for each component to serve best for its respective purpose. The system 
consists of a tissue-mimicking tube, which is designated to hold the capsule in a portion 
that is squeezed on both sides around the capsule edges to limit its motion. This 
squeezing is achieved with peristaltic rings, which are interchangeable to modify the 
change in diameter that occurs with muscle contractions. The tube is fastened on both 
sides to a chain looped around a motor-interfacing gear, which can be programmed for 
different speeds. Different programming can aid in mimicking different peristaltic 
motility conditions, while the lumen of the tube can be modified with excised tissue 
samples. Furthermore, the tube contains fluidic inlets on either side (one of which can 
function as an outlet), allowing for the implementation of different fluid profiles 
123 
 
 
 
described previously in this section. The use cases of such a system are for testing of 
features such as (a) sensor noise due to peristalsis, (b) capsule mechanical integrity, 
and (c) the supplemental effect of GI-simulating fluids on capsule integrity and 
sampling. While the system is limited regarding printable materials and motor 
robustness, there is potential to isolate relevant peristalsis conditions that can aid 
predicting the efficacy of targeting or sensing functions of the capsule. 
 
Figure 5-4 Peristalsis simulator, consisting mostly of 3D printed components. 
5.2.2 Anchoring Capsules 
 Because content transit through the GI tract is generally a temporary process, 
most capsule devices are designed for metric measurement as single time points along 
specific locations of the canal trajectory. However, some conditions would benefit 
more from continuous monitoring of a metric at the same location over extended time 
periods, such as ulcer progression in the stomach or inflammation in the colon [242]?
[244]. System implantation or retention is a function required for several devices 
intended for long-term residence in the GI tract, though it can be challenging to embed 
such functions into a capsule without clinical interventions such as surgery [245]. 
124 
 
 
 
 One of the challenges with adhesion in the GI tract is due to surface variation. 
For example, structures such as villi and microvilli, such as those depicted in Figure 
5-5a, are present along the lumen, which are critical for increasing intestinal surface 
area and therefore nutrient absorption into underlying blood vessels [246], [247]. After 
investigating various fabrication processes, one of my mentees, Aditya Nilacantan,  had 
3D-printed example villi surfaces for potential testing with capsule-based anchoring 
structures. Some work has been done on capsule-based adhesion mechanisms to the GI 
mucosa (Figure 5-5c), though we have also designed various microhooks, bioinspired 
by microneedles from common GI parasites, that we have also 3D-printed for facile 
integration with our capsule shells (Figure 5-5d) [231], [248]. Though the depicted 
concept would require appropriate spatial and mass distribution of electronic 
components, the microhooks could enable a passive approach to anchor the capsule to 
the mucosa for measuring a metric in the same location for multiple instances. We 
further envision, due to the ability to simultaneously and wirelessly transmit signals to 
our microelectronics platform, the active triggering of anchor actuation upon external 
command. Such advancements would be critical to establishing further levels of control 
to the system while in vivo, thereby expanding the time scales for measuring specific 
analytes of interest at target locations in the GI tract. 
125 
 
 
 
(a) (b)  
(c) (d)  
Figure 5-5 SEM imaging of (a) porcine (reproduced with permission from [246]) and (b) 3D-printed villi 
with an EnvisionTEC Perfactory 4. (c) Implantation capsule robot (reproduced with permission from 
[249]). (d) 3D-printed anchoring concept for passive capsule implantation; left inset: head of Taenia 
Solium Scolex (reproduced with permission from [248]), right inset: close-up of microhook mimic. 
5.2.3 Feedback-Triggered Therapeutic Release 
A primary purpose of biosensors in the clinical setting is to understand the state of a 
physiological metric and determine whether intervention is necessary to revert an 
unhealthy or pathological condition back to homeostasis for said metric. For example, 
patients with diabetes, after learning through a blood glucose sensor that they have 
high blood sugar, understand that they require insulin [250]. Figure 5-6 depicts the 
organs in or closely related to the GI tract, and lists various associated conditions that 
ingestible capsules could aid through feedback-triggered therapeutics [251]. 
126 
 
 
 
 
Figure 5-6 Physiology of GI affected regions, reflecting potentially relevant information accessible by 
capsule devices. 
 The same would be said for the detected analytes from the work in this 
dissertation, i.e. low levels of pancreatic trypsin or lipase resulting from conditions 
such as pancreatitis, pancreatic cancer, or cystic fibrosis would prompt a dose or 
supplement of these enzymes from an external source [1], [4], [150]. Because the 
platform is programmable, it is well within the system limits to use the acquired 
information to trigger another system to release enzyme supplements. This would 
require various conditions to be satisfied: (1) the information would need to be 
processed appropriately such that an algorithm for interpreting the film response would 
indicate an enzyme efficiency, and (2) the platform would need to be modified to allow 
control over a mechanism, using either the previous algorithm or patient-enabled from 
a button on the mobile app, for either actuating a drug-release switch or dissolving a 
reservoir-enclosing membrane [231]. 
127 
 
 
 
5.3 Conclusion 
 The GI tract is a complex environment that, while relatively well understood, 
continues to demonstrate itself as a hub for unknown physiological interactions that 
demand interrogation. Ingestible devices require a small form factor to reduce potential 
downstream risk to patient health, which, combined with the current state of 
microelectronics components and a lack of control inherent to content consumption, 
remain significant limitations to the range of achievable device functions. This 
dissertation addresses fabrication and integration strategies in embedding novel capsule 
functions by designing systems for (a) interfacing various biochemical reactions with 
electronic sensing modalities through biomaterial films and (b) targeted sampling of 
GI-simulating environments and their contents using pH-soluble coatings materials as 
system packaging. Disassociating electrical signal and system elements from 
interfacing with biomaterial reactions led to a platform design that, while useful for a 
variety of capacitive and/or film-based sensing challenges, was demonstrated here for 
detecting pancreatic enzyme activity and characterizing packaging integrity within the 
context of an integrated capsule. 
 The discussed designs, fabrication processes, and resulting assembled devices 
represent significant technological advances to bioelectronics interfacing and the field 
of ingestible technologies. The final capsule sets a precedent for GI targeting 
microsensors and the types of biomolecular species that can be detected, 
revolutionizing the capabilities of such devices for aiding healthcare paradigms from 
new perspectives and in low risk settings. The device and materials are relatively low 
cost, with facile interfacing with users (i.e. patients or clinicians) toward establishing 
128 
 
 
 
new goals in personalized medicine. The capsule serves additionally as a research tool, 
enabling the elucidation of how target physiological metrics, such as enzymes or bile 
acids, respond in situ to different stimuli and during varying conditions. Ultimately, the 
strategies implemented with this capsule platform for targeting sampling and sensing 
in GI environments facilitate future development in non-invasive GI diagnostics, both 
in academia regarding the manner in which molecular trends are observed during 
different pathologies, as well as clinically due to a reduction in the clinical interface 
required to obtain medically-relevant information. 
 
  
129 
 
 
 
6. Appendices 
6.1 Appendix A: Masks Used 
A.1 Microfluidic Impedance Sensing 
 
Figure 6-1 IDE sensors used in chapter 2. Finger width: 2 ?m, finger spacing: 4 ?m, finger length: 1 mm, 
number of fingers per electrode: 65.  
 
130 
 
 
 
A.2 First Capsule Mask for Capacitive Sensing 
  
Figure 6-2 IDE sensor patterns used in chapter 3 to measure integrity of the packaging coatings. (a) 
Complete photomask with 10 die for each 4x1 sensor die, each with different finger geometries: 25, 50, 
and 75 ?m width and spacing and 44, 22, and 14 fingers per electrode, respectively. Finger length: 2 mm. 
 
131 
 
 
 
A.3 Sensors used for SPA 
 
Figure 6-3 IDE sensor patterns used in chapters 3 and 4 for SPAs. Mask Design: 124 Sensor Output, 62 
each of long (length=3 mm) /short (1.5 mm) versions. Number of fingers per electrodes = 40. 
 
 
 
 
 
 
 
 
 
 
 
 
132 
 
 
 
6.2 Appendix B: COMSOL Simulation and Results 
 To reduce computational processing requirements, I reduced the sensor to a 
two-finger unit rather than 146 fingers, used earlier in this chapter. One finger is 
denoted as working electrode while the other is the ground, and the material properties 
and dimensions used are listed in Table 6-1. A cross-sectioned diagram of our model 
and the conditions I used are depicted below in a, electrodes are simulated with either 
no film, a 0.5 ?m thick (to represent partially digested), and 1 ?m thick (to represent 
undigested) films. For comparison, the capacitance of the sensors was measured using 
LCR meter (Agilent E4980A) controlled with LabVIEW. Measurements were carried 
out sequentially in air, in phosphate buffer (PB; pH 7.5), and finally in 1 mg/mL bovine 
pancreatic trypsin in PB at 37?C (physiological temperature) under constant stirring. 
Sensors tested were either coated in gelatin or uncoated. AC signals were swept from 
100 kHz to 1 MHz at 50 mV in every 90 seconds. I observed films before and after 
each experiment and compared extracted parallel capacitance experimental results to 
COMSOL Multiphysics simulation results. Direct DC capacitance responses were 
calculated for films with a 100 ?m average thickness or 0 ?m (no film) surrounded by 
either buffer or air. Simulation capacitance was obtained for only two fingers in the 
sensor, but was scaled up accordingly for IDEs [252].  
Table 6-1 Parameters used for the COMSOL simulation. 
Material Glass Gelatin (film) Pbs Gold (electrodes) 
(substrate) (buffer) 
Conductivity (s/m) 1.0e-14 7.0e-5 1.52 44.2e6 
Dimension  1 mm x 6 ?m 1 mm x 6 ?m 1 mm x 6 1 mm x 1 ?m x 200 nm 
(width x depth x x 0.4 ?m x 1-0.5 ?m ?m x 10 ?m (x2) (4 ?m spacing) 
height) 
 
133 
 
 
 
 Figure 6-4 (b) demonstrates a comparison of capacitance signals for both 
experimental and simulation results. As indicated by the y-axes, the capacitance 
magnitudes from the experiment are slightly offset from the simulation results, though 
the trends are similar. The decrease in capacitance for the film-coated sensors from air 
to buffer are interesting but not unexpected, as it likely reflects charge dissipation from 
the film, which is at the surface of the electrode, to the buffer due to its electrical 
conductivity (air: 5.5E-15 S/M compared to PBS: 1.52 S/M). The step change in 
capacitance due to buffer contact to the electrodes is precedented however, and 
indicates the potential for monitoring capacitance alone to measure film digestion 
rather than total impedance. This result becomes reflected further in the next few 
chapters, both for monitoring film digestion and sampling of environmental buffer. 
134 
 
 
 
(a)  
 (b)  
Figure 6-4 (a) Cross-section of simulation model (top to bottom: medium, film, sensor, wafer), (b) mean 
capacitance from experimental degradation of film over IDE sensor (left y-axis) and simulation (right y-
axis) results. Between electrode fingers, electric field intensity increases significantly in the presence of 
gelatin. Electric field height decreases significantly, the magnitude remains the same, and lateral electric 
field distribution is altered. 
 
6.4 Appendix C: MATLAB Analysis Code and GUI 
 This MATLAB code automatically initializes a GUI, depicted in Figure 6-5, 
that enables input of up to four data sequences over time, with customization over time 
and percent change in capacitance (x- and y-axes, respectively) ranges, applied 
smoothing filters, and legend labels. The sequences used must be in a .csv file format 
with data arrangement such as that depicted in Figure 6-6. 
 
 
135 
 
 
 
 
function varargout = multi_data_sequences_5_31_19(varargin) 
% MULTI_DATA_SEQUENCES_5_31_19 MATLAB code for 
multi_data_sequences_5_31_19.fig 
%      MULTI_DATA_SEQUENCES_5_31_19, by itself, creates a new 
MULTI_DATA_SEQUENCES_5_31_19 or raises the existing 
%      singleton*. 
% 
%      H = MULTI_DATA_SEQUENCES_5_31_19 returns the handle to a new 
MULTI_DATA_SEQUENCES_5_31_19 or the handle to 
%      the existing singleton*. 
% 
%      
MULTI_DATA_SEQUENCES_5_31_19('CALLBACK',hObject,eventData,handles,..
.) calls the local 
%      function named CALLBACK in MULTI_DATA_SEQUENCES_5_31_19.M 
with the given input arguments. 
% 
%      MULTI_DATA_SEQUENCES_5_31_19('Property','Value',...) creates 
a new MULTI_DATA_SEQUENCES_5_31_19 or raises the 
%      existing singleton*.  Starting from the left, property value 
pairs are 
%      applied to the GUI before 
multi_data_sequences_5_31_19_OpeningFcn gets called.  An 
%      unrecognized property name or invalid value makes property 
application 
%      stop.  All inputs are passed to 
multi_data_sequences_5_31_19_OpeningFcn via varargin. 
% 
%      *See GUI Options on GUIDE's Tools menu.  Choose "GUI allows 
only one 
%      instance to run (singleton)". 
% 
% See also: GUIDE, GUIDATA, GUIHANDLES 
  
% Edit the above text to modify the response to help 
multi_data_sequences_5_31_19 
  
% Last Modified by GUIDE v2.5 05-Jun-2019 15:50:08 
  
% Begin initialization code - DO NOT EDIT 
gui_Singleton = 1; 
gui_State = struct('gui_Name',       mfilename, ... 
                   'gui_Singleton',  gui_Singleton, ... 
                   'gui_OpeningFcn', 
@multi_data_sequences_5_31_19_OpeningFcn, ... 
                   'gui_OutputFcn',  
@multi_data_sequences_5_31_19_OutputFcn, ... 
                   'gui_LayoutFcn',  [] , ... 
                   'gui_Callback',   []); 
if nargin && ischar(varargin{1}) 
    gui_State.gui_Callback = str2func(varargin{1}); 
end 
  
if nargout 
136 
 
 
 
    [varargout{1:nargout}] = gui_mainfcn(gui_State, varargin{:}); 
else 
    gui_mainfcn(gui_State, varargin{:}); 
end 
% End initialization code - DO NOT EDIT 
  
  
% --- Executes just before multi_data_sequences_5_31_19 is made 
visible. 
function multi_data_sequences_5_31_19_OpeningFcn(hObject, eventdata, 
handles, varargin) 
  
handles.output = hObject; 
guidata(hObject, handles); 
  
% --- Outputs from this function are returned to the command line. 
function varargout = multi_data_sequences_5_31_19_OutputFcn(hObject, 
eventdata, handles)  
varargout{1} = handles.output; 
  
  
%_________________Set Value Objects for Plots_____________________% 
  
%_______________Filter Setting_______________% 
  
function edit1_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit1_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
% --- Executes on button press in pushbutton9. 
function pushbutton9_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
s1 = get(handles.edit1,'String');  
handles.filter=str2double(s1); 
guidata(hObject, handles); 
  
  
%_____________________End Time 1 Setting____________________% 
  
function edit5_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit5_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
137 
 
 
 
% --- Executes on button press in pushbutton11. 
function pushbutton11_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
et1 = get(handles.edit5,'String'); 
endtime1=str2double(et1); 
guidata(hObject, handles); 
axes(handles.axes1); 
xlim([0 endtime1]) 
  
  
%_____________________End Time 2 Setting____________________% 
  
function edit6_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit6_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
% --- Executes on button press in pushbutton12. 
function pushbutton12_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
et2 = get(handles.edit6,'String'); 
endtime2=str2double(et2); 
guidata(hObject, handles); 
axes(handles.axes2); 
xlim([0 endtime2]) 
  
  
%_____________________End Time 3 Setting____________________% 
  
function edit7_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit7_CreateFcn(hObject, eventdata, handles) 
  
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
  
% --- Executes on button press in pushbutton13. 
function pushbutton13_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
et3 = get(handles.edit7,'String'); 
endtime3=str2double(et3); 
guidata(hObject, handles); 
axes(handles.axes3); 
xlim([0 endtime3]) 
  
138 
 
 
 
  
%_____________________End Time 4 Setting____________________% 
  
  
function edit8_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit8_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
% --- Executes on button press in pushbutton14. 
function pushbutton14_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
et4 = get(handles.edit8,'String'); 
endtime4=str2double(et4); 
guidata(hObject, handles); 
axes(handles.axes4); 
xlim([0 endtime4]) 
  
  
  
  
% ___________________________PLOTTING DATA_______________________% 
  
% --- Executes on button press in pushbutton1. 
function pushbutton1_Callback(hObject, eventdata, handles) 
  
global filename1 
global pathname 
global currentfolder 
  
[filename1, pathname] = uigetfile({'C:\George\Capsule 
Experiments\BLE\Calibration\*.csv'}) 
currentfolder = fullfile(pathname, filename1); 
if isequal(filename1,0) || isequal(pathname,0) 
    return 
end 
  
fileID = fopen(fullfile(pathname, filename1)); 
  
temp = fscanf(fileID,'%d',[8 Inf]); 
handles.fileData = temp'; 
guidata(hObject, handles); 
  
  
%____________________________Plot 1 Stuff_________________________% 
  
%_____________below this, actual time_____________________% 
%path=('C:\George\'); 
path=pathname; 
139 
 
 
 
file_path=strcat(path,filename1); 
class(file_path); 
Time = GetFileTime(file_path); 
Time_1=struct2cell(Time); 
finish=Time_1(3); 
start=strsplit(filename1,"-"); 
start_date=str2double(start(5)); 
%start_hour=str2double(start(6)); 
%start_min=str2double(start(7)); 
global name1 
  
table1 = readtable(filename1); 
sensor1=table2array(table1(:,2)); 
  
end_date=start_date; 
end_totaltime=table2array(table1(:,1)); 
  
timex1=cell2mat(end_totaltime(1,1)); 
start_time=strsplit(timex1,":"); 
start_hour=str2double(start_time(1)); 
start_min=str2double(start_time(2)); 
timex=cell2mat(end_totaltime(length(end_totaltime),1)); 
end_time=strsplit(timex,":"); 
end_hour=str2double(end_time(1)); 
end_min=str2double(end_time(2)); 
  
cap_val_1=[]; 
for j=1:length(sensor1) 
    volt_val_1(j,:) = 1/((4095/3.3)/(sensor1(j)))-0.05; 
    %cap_val_1(j,:) = ((volt_val_1(j,:)-volt_val_1(1))/0.0031); 
    cap_val_1(j,:) = ((volt_val_1(j,:)-2.5325)/0.0031); 
    delta_cap_1(j,:) = abs(cap_val_1(j,:)-
cap_val_1(1,:))*100/abs(cap_val_1(1,:));     
end 
total_cap_val_1=cap_val_1; 
total_delta_cap_val_1=delta_cap_1; 
  
%__________________________Setting up time_______________________% 
N= length(sensor1); 
Total_time =  (end_date - start_date) * 24 * 60 * 60 + (end_hour - 
start_hour)* 60 * 60 + (end_min - start_min) * 60;  % Calculate 
total time in seconds 
Time_step = Total_time/(N-1); % average step time in seconds between 
two consecutive runs 
Time_s = 0: Time_step: Total_time; % create an array called time 
Time_m = Time_s/(60); % row array of time in hours 
m=1; 
  
%_________________________ plot 1 stuff_________________________% 
filter = handles.filter; 
axes(handles.axes1); 
handles.h1=plot(handles.axes1,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'ko-
140 
 
 
 
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
mean(total_delta_cap_val_1(8:24)) 
std(total_delta_cap_val_1(8:24)) 
total_delta_cap_val_1=total_delta_cap_val_1.'; 
%p=polyfit(Time_m(1:2000),total_delta_cap_val_1(1:2000),1); %for 
%publications, 2000 points = 33 min 
et1 = get(handles.edit5,'String'); 
endtime1=str2double(et1); 
endtime1p=endtime1*60; 
p=polyfit(Time_m(1:endtime1p),total_delta_cap_val_1(1:endtime1p),1); 
xp = linspace(0,endtime1); 
yp = polyval(p,xp); 
plot(handles.axes1,xp,yp+2,'k-','LineWidth',2) %display when want 
%trendline 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 150]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
% Place equation in upper left of graph. 
xl = xlim; 
yl = ylim; 
xt = 0.05 * (xl(2)-xl(1)) + xl(1); 
yt = 0.90 * (yl(2)-yl(1)) + yl(1); 
caption = sprintf('y = %f * x + %f', p(1), p(2)); 
text(handles.axes1,xt, yt, caption, 'FontSize', 10, 'Color', 'k'); 
%%display when wanting equation 
  
[filepath,name1,ext] = fileparts(filename1); 
set(handles.text2, 'String', name1); 
set(gcf,'PaperPositionMode','auto') 
print([num2str(name1)],'-dtiffn','-r0') 
  
%_________________________group plot stuff________________________% 
axes(handles.axes4); 
handles.h1=plot(handles.axes4,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'ko-
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 100]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
%_______________________END DATA 1_______________________________% 
  
  
  
141 
 
 
 
  
  
% --- Executes on button press in pushbutton2. 
function pushbutton2_Callback(hObject, eventdata, handles) 
  
global filename2 
global pathname2 
global currentfolder2 
  
[filename2, pathname2] = uigetfile({'C:\George\Capsule 
Experiments\BLE\Calibration\*.csv'}) 
currentfolder2 = fullfile(pathname2, filename2); 
if isequal(filename2,0) || isequal(pathname2,0) 
    return 
end 
  
fileID = fopen(fullfile(pathname2, filename2)); 
  
temp = fscanf(fileID,'%d',[8 Inf]); 
handles.fileData = temp'; 
guidata(hObject, handles); 
  
%_________________________Plot 2 Stuff____________________________% 
  
%_____________below this, actual time_____________________% 
%path=('C:\George\'); 
path=pathname2; 
file_path=strcat(path,filename2); 
class(file_path); 
Time = GetFileTime(file_path); 
Time_1=struct2cell(Time); 
finish=Time_1(3); 
start=strsplit(filename2,"-"); 
start_date=str2double(start(5)); 
%start_hour=str2double(start(6)); 
%start_min=str2double(start(7)); 
global name2 
  
table1 = readtable(filename2); 
sensor1=table2array(table1(:,2)); 
  
end_date=start_date; 
end_totaltime=table2array(table1(:,1)); 
  
timex1=cell2mat(end_totaltime(1,1)); 
start_time=strsplit(timex1,":"); 
start_hour=str2double(start_time(1)); 
start_min=str2double(start_time(2)); 
timex=cell2mat(end_totaltime(length(end_totaltime),1)); 
end_time=strsplit(timex,":"); 
end_hour=str2double(end_time(1)); 
end_min=str2double(end_time(2)); 
  
cap_val_1=[]; 
142 
 
 
 
for j=1:length(sensor1) 
    volt_val_1(j,:) = 1/((4095/3.3)/(sensor1(j)))-0.05; 
    %cap_val_1(j,:) = ((volt_val_1(j,:)-volt_val_1(1))/0.0031); 
    cap_val_1(j,:) = ((volt_val_1(j,:)-2.5325)/0.0031); 
    delta_cap_1(j,:) = abs(cap_val_1(j,:)-
cap_val_1(1,:))*100/abs(cap_val_1(1,:));     
end 
  
total_cap_val_1=cap_val_1; 
total_delta_cap_val_1=delta_cap_1; 
  
%_________________________Setting up time__________________________% 
N= length(sensor1); 
Total_time =  (end_date - start_date) * 24 * 60 * 60 + (end_hour - 
start_hour)* 60 * 60 + (end_min - start_min) * 60;  % Calculate 
total time in seconds 
Time_step = Total_time/(N-1); % average step time in seconds between 
two consecutive runs 
Time_s = 0: Time_step: Total_time; % create an array called time 
Time_m = Time_s/(60); % row array of time in hours 
m=1; 
  
%__________________________ plot 2 stuff__________________________% 
filter = handles.filter; 
axes(handles.axes2); 
handles.h2=plot(handles.axes2,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'kd-
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
total_delta_cap_val_1=total_delta_cap_val_1.'; 
%p=polyfit(Time_m(1:2000),total_delta_cap_val_1(1:2000),1); %for 
%publications, 2000 points = 33 min 
et1 = get(handles.edit6,'String'); 
endtime1=str2double(et1); 
endtime1p=endtime1*60; 
p=polyfit(Time_m(1:endtime1p),total_delta_cap_val_1(1:endtime1p),1); 
xp = linspace(0,endtime1); 
yp = polyval(p,xp); 
plot(handles.axes2,xp,yp+2,'k-','LineWidth',2) 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 150]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
% Place equation in upper left of graph. 
xl = xlim; 
yl = ylim; 
xt = 0.05 * (xl(2)-xl(1)) + xl(1); 
yt = 0.90 * (yl(2)-yl(1)) + yl(1); 
caption = sprintf('y = %f * x + %f', p(1), p(2)); 
text(handles.axes2,xt, yt, caption, 'FontSize', 10, 'Color', 'k'); 
143 
 
 
 
  
[filepath,name2,ext] = fileparts(filename2); 
set(handles.text3, 'String', name2); 
set(gcf,'PaperPositionMode','auto') 
print([num2str(name2)],'-dtiffn','-r0') 
  
axes(handles.axes4); 
handles.h2=plot(handles.axes4,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'kd-
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 100]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
%___________________________END DATA 2_____________________________% 
 
  
% --- Executes on button press in pushbutton3. 
function pushbutton3_Callback(hObject, eventdata, handles) 
  
global filename3 
global pathname3 
global currentfolder 
  
[filename3, pathname3] = uigetfile({'C:\George\Capsule 
Experiments\BLE\Calibration\*.csv'}) 
currentfolder = fullfile(pathname3, filename3); 
if isequal(filename3,0) || isequal(pathname3,0) 
    return 
end 
  
fileID = fopen(fullfile(pathname3, filename3)); 
  
temp = fscanf(fileID,'%d',[8 Inf]); 
handles.fileData = temp'; 
guidata(hObject, handles); 
  
%________________________Plot 3 Stuff_____________________________% 
  
%_____________below this, actual time_____________________% 
%path=('C:\George\'); 
path=pathname3; 
file_path=strcat(path,filename3); 
class(file_path); 
Time = GetFileTime(file_path); 
Time_1=struct2cell(Time); 
finish=Time_1(3); 
start=strsplit(filename3,"-"); 
start_date=str2double(start(5)); 
144 
 
 
 
%start_hour=str2double(start(6)); 
%start_min=str2double(start(7)); 
global name3 
  
table1 = readtable(filename3); 
sensor1=table2array(table1(:,2)); 
  
end_date=start_date; 
end_totaltime=table2array(table1(:,1)); 
  
timex1=cell2mat(end_totaltime(1,1)); 
start_time=strsplit(timex1,":"); 
start_hour=str2double(start_time(1)); 
start_min=str2double(start_time(2)); 
timex=cell2mat(end_totaltime(length(end_totaltime),1)); 
end_time=strsplit(timex,":"); 
end_hour=str2double(end_time(1)); 
end_min=str2double(end_time(2)); 
  
cap_val_1=[]; 
for j=1:length(sensor1) 
    volt_val_1(j,:) = 1/((4095/3.3)/(sensor1(j)))-0.05; 
    %cap_val_1(j,:) = ((volt_val_1(j,:)-volt_val_1(1))/0.0031); 
    cap_val_1(j,:) = ((volt_val_1(j,:)-2.5325)/0.0031); 
    delta_cap_1(j,:) = abs(cap_val_1(j,:)-
cap_val_1(1,:))*100/abs(cap_val_1(1,:));     
end 
total_cap_val_1=cap_val_1; 
total_delta_cap_val_1=delta_cap_1; 
  
%__________________________Setting up time________________________% 
N= length(sensor1); 
Total_time =  (end_date - start_date) * 24 * 60 * 60 + (end_hour - 
start_hour)* 60 * 60 + (end_min - start_min) * 60;  % Calculate 
total time in seconds 
Time_step = Total_time/(N-1); % average step time in seconds between 
two consecutive runs 
Time_s = 0: Time_step: Total_time; % create an array called time 
Time_m = Time_s/(60); % row array of time in hours 
m=1; 
%_________________________ plot 3 stuff____________________________% 
filter = handles.filter; 
axes(handles.axes3); 
handles.h3=plot(handles.axes3,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'k*-
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
total_delta_cap_val_1=total_delta_cap_val_1.'; 
%p=polyfit(Time_m(1:2000),total_delta_cap_val_1(1:2000),1); %for 
%publications, 2000 points = 33 min 
et1 = get(handles.edit7,'String'); 
endtime1=str2double(et1); 
endtime1p=endtime1*60; 
145 
 
 
 
p=polyfit(Time_m(1:endtime1p),total_delta_cap_val_1(1:endtime1p),1); 
xp = linspace(0,endtime1); 
yp = polyval(p,xp); 
plot(handles.axes3,xp,yp+2,'k-','LineWidth',2) 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 50]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
% Place equation in upper left of graph. 
xl = xlim; 
yl = ylim; 
xt = 0.05 * (xl(2)-xl(1)) + xl(1); 
yt = 0.90 * (yl(2)-yl(1)) + yl(1); 
caption = sprintf('y = %f * x + %f', p(1), p(2)); 
text(handles.axes3,xt, yt, caption, 'FontSize', 10, 'Color', 'k'); 
  
[filepath,name3,ext] = fileparts(filename3); 
set(handles.text4, 'String', name3); 
set(gcf,'PaperPositionMode','auto') 
print([num2str(name3)],'-dtiffn','-r0') 
  
axes(handles.axes4); 
handles.h3=plot(handles.axes4,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'k*-
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 150]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
%_______________________END DATA 3________________________________% 
  
% --- Executes on button press in pushbutton4. 
function pushbutton4_Callback(hObject, eventdata, handles) 
cla(handles.axes1,'reset'); 
fclose('all'); 
  
% --- Executes on button press in pushbutton5. 
function pushbutton5_Callback(hObject, eventdata, handles) 
cla(handles.axes2,'reset'); 
fclose('all'); 
  
% --- Executes on button press in pushbutton6. 
function pushbutton6_Callback(hObject, eventdata, handles) 
cla(handles.axes3,'reset'); 
fclose('all'); 
  
  
146 
 
 
 
% --- Executes on button press in pushbutton7. 
function pushbutton7_Callback(hObject, eventdata, handles) 
cla(handles.axes4,'reset'); 
fclose('all'); 
  
function edit2_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit2_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
function edit3_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit3_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
 
function edit4_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit4_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
% --- Executes on button press in pushbutton10. 
function pushbutton10_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
  
l1 = get(handles.edit2,'String');  
l2 = get(handles.edit3,'String');  
l3 = get(handles.edit4,'String');  
guidata(hObject, handles); 
  
axes(handles.axes4); 
legend(l1,l2, l3,'Location','northwest') 
legend boxoff 
  
  
function edit9_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit9_CreateFcn(hObject, eventdata, handles) 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
147 
 
 
 
  
  
% --- Executes on button press in pushbutton15. 
function pushbutton15_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
my1 = get(handles.edit9,'String'); 
maxy1=str2double(my1); 
guidata(hObject, handles); 
axes(handles.axes4); 
ylim([0 maxy1]) 
  
%________________________Additional Plot Stuff____________________% 
  
% --- Executes on button press in pushbutton16. 
function pushbutton16_Callback(hObject, eventdata, handles) 
  
global filename1 
global pathname 
global currentfolder 
  
[filename1, pathname] = uigetfile({'C:\George\Capsule 
Experiments\BLE\Calibration\*.csv'}) 
currentfolder = fullfile(pathname, filename1); 
if isequal(filename1,0) || isequal(pathname,0) 
    return 
end 
  
fileID = fopen(fullfile(pathname, filename1)); 
  
temp = fscanf(fileID,'%d',[8 Inf]); 
handles.fileData = temp'; 
guidata(hObject, handles); 
  
%_____________________Additional Plot Stuff________________________% 
  
%_____________below this, actual time_____________________% 
%path=('C:\George\'); 
path=pathname; 
file_path=strcat(path,filename1); 
class(file_path); 
Time = GetFileTime(file_path); 
Time_1=struct2cell(Time); 
finish=Time_1(3); 
start=strsplit(filename1,"-"); 
start_date=str2double(start(5)); 
start_hour=str2double(start(6)); 
start_min=str2double(start(7)); 
global name1 
  
table1 = readtable(filename1); 
sensor1=table2array(table1(:,2)); 
  
end_date=start_date; 
end_totaltime=table2array(table1(:,1)); 
148 
 
 
 
timex=cell2mat(end_totaltime(length(end_totaltime),1)); 
end_time=strsplit(timex,":"); 
end_hour=str2double(end_time(1)); 
end_min=str2double(end_time(2)); 
  
cap_val_1=[]; 
for j=1:length(sensor1) 
    volt_val_1(j,:) = 1/((4095/3.3)/(sensor1(j)))-0.05; 
    %cap_val_1(j,:) = ((volt_val_1(j,:)-volt_val_1(1))/0.0031); 
    cap_val_1(j,:) = ((volt_val_1(j,:)-2.5325)/0.0031); 
    delta_cap_1(j,:) = abs(cap_val_1(j,:)-
cap_val_1(1,:))*100/abs(cap_val_1(1,:));     
end 
  
total_cap_val_1=cap_val_1; 
total_delta_cap_val_1=delta_cap_1; 
  
%________________________Setting up time__________________________% 
  
N= length(sensor1); 
Total_time =  (end_date - start_date) * 24 * 60 * 60 + (end_hour - 
start_hour)* 60 * 60 + (end_min - start_min) * 60;  % Calculate 
total time in seconds 
Time_step = Total_time/(N-1); % average step time in seconds between 
two consecutive runs 
Time_s = 0: Time_step: Total_time; % create an array called time 
Time_m = Time_s/(60); % row array of time in hours 
m=1; 
  
[filepath,name1,ext] = fileparts(filename1); 
set(handles.text6, 'String', name1); 
set(gcf,'PaperPositionMode','auto') 
print([num2str(name1)],'-dtiffn','-r0') 
  
%______________________group plot stuff____________________________% 
filter = handles.filter; 
axes(handles.axes4); 
handles.h4=plot(handles.axes4,Time_m(m:length(total_delta_cap_val_1)
),smooth(total_delta_cap_val_1(1:length(total_delta_cap_val_1)),filt
er),'k--
','MarkerSize',5,'MarkerIndices',1:75:length(total_delta_cap_val_1))
; 
hold on 
xlabel('Time (min)', 'FontSize',12,'FontWeight','bold') 
ylabel('\DeltaC (%)', 'FontSize',12,'FontWeight','bold') 
%ylim([0 100]) 
%xlim([0 60]) 
set(gca,'fontsize',12,'FontWeight','bold','box','off') 
ax = gca; 
%____________________END Additional Data__________________________% 
  
function edit10_Callback(hObject, eventdata, handles) 
  
% --- Executes during object creation, after setting all properties. 
function edit10_CreateFcn(hObject, eventdata, handles) 
149 
 
 
 
if ispc && isequal(get(hObject,'BackgroundColor'), 
get(0,'defaultUicontrolBackgroundColor')) 
    set(hObject,'BackgroundColor','white'); 
end 
  
  
% --- Executes on button press in pushbutton18. 
function pushbutton18_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
  
l1 = get(handles.edit2,'String');  
l2 = get(handles.edit3,'String');  
l3 = get(handles.edit4,'String');  
l4 = get(handles.edit10,'String');  
guidata(hObject, handles); 
  
axes(handles.axes4); 
legend(l1,l2, l3, l4,'Location','northwest') 
legend boxoff 
  
  
% --- Executes on button press in pushbutton19. 
function pushbutton19_Callback(hObject, eventdata, handles) 
handles = guidata(hObject); 
guidata(hObject, handles); 
h1 = get(handles.h1);  
h2 = get(handles.h2);  
h3 = get(handles.h3);  
h4 = get(handles.h4);  
refreshdata(h1); 
refreshdata(h2); 
refreshdata(h3); 
refreshdata(h4); 
 
150 
 
 
 
 
Figure 6-5 MATLAB GUI. Up to four data sequences can be plotted, three of which can be plotted 
individually on the right while all are overlaid in the plot on the bottom left. Features that can be modified 
directly in the GUI are smoothing filters, axes limits, and legend labels. 
 
 
Figure 6-6 Depiction of data format in .csv file. This format is currently programmed into the app, while 
the algorithms to process the capacitance data are embedded in the MATLAB code above. 
151 
 
 
 
7. Bibliography 
[1] D. M. Goldberg and K. G. Wormsley, ?The interrelationships of pancreatic 
enzymes in human duodenal aspirate,? Gut, vol. 11, no. 10, pp. 859?866, 1970. 
[2] R. C. Hartley, E. E. Gambill, G. W. Engstrom, and W. H. J. Summerskill, 
?Pancreatic exocrine function,? Am. J. Dig. Dis., vol. 11, no. 1, pp. 27?39, 
1966. 
[3] M. E. Haupt, M. J. Kwasny, M. S. Schechter, and S. A. McColley, ?Pancreatic 
Enzyme Replacement Therapy Dosing and Nutritional Outcomes in Children 
with Cystic Fibrosis,? J. Pediatr., vol. 164, no. 5, pp. 1110-1115.e1, 2014. 
[4] J. G. L. II and P. V Draganov, ?Pancreatic function testing: Here to stay for the 
21st century,? World J. Gastroenterol., vol. 14, no. 20, pp. 3149?3158, 2008. 
[5] K. Strimbu and J. A. Tavel, ?What are biomarkers?,? Curr. Opin. HIV AIDS, 
vol. 5, no. 6, pp. 463?6, Nov. 2010. 
[6] S.-L. Liang and D. W. Chan, ?Enzymes and related proteins as cancer 
biomarkers: a proteomic approach.,? Clin. Chim. Acta., vol. 381, no. 1, pp. 93?
7, May 2007. 
[7] S. J. Pandol, ?The Exocrine Pancreas,? Morgan Claypool Life Sci., 2010. 
[8] J. Wang et al., ?Circulating microRNAs in Pancreatic Juice as Candidate 
Biomarkers of Pancreatic Cancer,? J. Cancer, vol. 5, no. 8, pp. 696?705, 2014. 
[9] Z. Altintas, S. S. Kallempudi, and Y. Gurbuz, ?Gold nanoparticle modified 
capacitive sensor platform for multiple marker detection,? Talanta, vol. 118, 
no. Supplement C, pp. 270?276, 2014. 
[10] D. Friedel, R. Modayil, and S. Stavropoulos, ?Colon Capsule Endoscopy: 
152 
 
 
 
Review and Perspectives,? Gastroenterol. Res. Pract., vol. 2016, p. 9643162, 
2016. 
[11] S. Yim, E. Gultepe, D. H. Gracias, and M. Sitti, ?Biopsy using a Magnetic 
Capsule Endoscope Carrying, Releasing, and Retrieving Untethered 
Microgrippers,? IEEE Trans. Biomed. Eng., vol. 61, no. 2, pp. 513?521, 2014. 
[12] J. Z. Ou, C. K. Yao, A. Rotbart, J. G. Muir, P. R. Gibson, and K. Kalantar-
zadeh, ?Human intestinal gas measurement systems: in vitro fermentation and 
gas capsules,? Trends Biotechnol., vol. 33, no. 4, pp. 208?213, 2015. 
[13] J. Z. Ou et al., ?Potential of in vivo real-time gastric gas profiling: a pilot 
evaluation of heat-stress and modulating dietary cinnamon effect in an animal 
model,? Sci. Rep., vol. 6, p. 33387, 2016. 
[14] S. E. Thompson and S. Parthasarathy, ?Moore?s law: the future of Si 
microelectronics,? Mater. Today, vol. 9, no. 6, pp. 20?25, Jun. 2006. 
[15] C. Cordon, System Theories: An Overview of Various System Theories and Its 
Application in Healthcare, vol. 2. 2013. 
[16] ?Effects of pH (Introduction to Enzymes).? [Online]. Available: 
http://www.worthington-biochem.com/introbiochem/effectspH.html. 
[Accessed: 16-Apr-2019]. 
[17] N. (Neil) Hopkinson, R. J. M. Hague, and P. M. Dickens, Rapid 
manufacturing?: an industrial revolution for the digital age. John Wiley, 2006. 
[18] M. Muskovich and C. J. Bettinger, ?Biomaterials-Based Electronics: Polymers 
and Interfaces for Biology and Medicine,? Adv. Healthc. Mater., vol. 1, no. 3, 
pp. 248?266, 2012. 
153 
 
 
 
[19] C. J. Bettinger and Z. Bao, ?Biomaterials-based organic electronic devices,? 
Polym. Int., vol. 59, no. 5, p. n/a-n/a, May 2010. 
[20] G. Voskerician et al., ?Biocompatibility and biofouling of MEMS drug 
delivery devices,? Biomaterials, vol. 24, no. 11, pp. 1959?1967, Jul. 2003. 
[21] M. J. Madou, From MEMS to Bio-MEMS and Bio-NEMS: Manufacturing 
Techniques and Applications. CRC Press, 2011. 
[22] L. H. Nielsen et al., ?pH-triggered drug release from biodegradable microwells 
for oral drug delivery,? Biomed. Microdevices, vol. 17, no. 3, p. 55, 2015. 
[23] E. T. Cole et al., ?Enteric coated HPMC capsules designed to achieve intestinal 
targeting,? Int. J. Pharm., vol. 231, no. 1, pp. 83?95, 2002. 
[24] A. Telefoncu, E. Dlnckaya, and K. D. Vorlop, ?Preparation and 
characterization of pancreatic lipase immobilized in Eudragit-matrix,? Appl. 
Biochem. Biotechnol., vol. 26, no. 3, pp. 311?317, Dec. 1990. 
[25] A. Touhami, ?Biosensors and Nanobiosensors: Design and Applications,? 
2014, pp. 374?403. 
[26] N. Bhalla, P. Jolly, N. Formisano, and P. Estrela, ?Introduction to biosensors,? 
Essays Biochem., vol. 60, no. 1, p. 1, 2016. 
[27] S. P. Mohanty and E. Kougianos, ?Biosensors: a tutorial review,? IEEE 
Potentials, vol. 25, no. 2, pp. 35?40, Mar. 2006. 
[28] D. R. Th?venot, K. Toth, R. A. Durst, and G. S. Wilson, ?Electrochemical 
biosensors: recommended definitions and classification,? Biosens. 
Bioelectron., vol. 16, no. 1?2, pp. 121?131, Jan. 2001. 
[29] S. Patel, R. Nanda, S. Sahoo, and E. Mohapatra, ?Biosensors in Health Care: 
154 
 
 
 
The Milestones Achieved in Their Development towards Lab-on-Chip-
Analysis,? Biochem. Res. Int., vol. 2016, pp. 1?12, Mar. 2016. 
[30] M. Chakraborty and M. Saleem J Hashmi, ?An Overview of Biosensors and 
Devices,? in Reference Module in Materials Science and Materials 
Engineering, 2017. 
[31] D. Grieshaber, R. MacKenzie, J. V?r?s, and E. Reimhult, ?Electrochemical 
Biosensors - Sensor Principles and Architectures,? Sensors (Basel)., vol. 8, no. 
3, p. 1400, 2008. 
[32] N. Elgrishi, K. J. Rountree, B. D. McCarthy, E. S. Rountree, T. T. Eisenhart, 
and J. L. Dempsey, ?A Practical Beginner?s Guide to Cyclic Voltammetry,? J. 
Chem. Educ., vol. 95, no. 2, pp. 197?206, Feb. 2018. 
[33] J. Wang, Analytical electrochemistry. Wiley-VCH, 2006. 
[34] R. van den Hurk and S. Evoy, ?A Review of Membrane-Based Biosensors for 
Pathogen Detection.,? Sensors (Basel)., vol. 15, no. 6, pp. 14045?78, Jun. 
2015. 
[35] N. T. Thet, ?Modified tethered bilayer lipid membranes for detection of 
pathogenic bacterial toxins and characterization of ion channels.? 2010. 
[36] K. Nakamura, T. Tanaka, and K. Takeo, ?Characterization of protein binding 
to a nitrocellulose membrane.,? SEIBUTSU BUTSURI KAGAKU, vol. 33, no. 
6, pp. 293?303, Dec. 1989. 
[37] X. Luo and J. J. Davis, ?Electrical biosensors and the label free detection of 
protein disease biomarkers,? Chem. Soc. Rev., vol. 42, no. 13, p. 5944, Jun. 
2013. 
155 
 
 
 
[38] R. A. Dorledo de Faria, L. G. Dias Heneine, T. Matencio, and Y. Messaddeq, 
?Faradaic and non-faradaic electrochemical impedance spectroscopy as 
transduction techniques for sensing applications,? Int. J. Biosens. Bioelectron., 
vol. 5, no. 1, Feb. 2019. 
[39] E. Katz and I. Willner, ?Probing Biomolecular Interactions at Conductive and 
Semiconductive Surfaces by Impedance Spectroscopy: Routes to Impedimetric 
Immunosensors, DNA-Sensors, and Enzyme Biosensors,? Electroanalysis, vol. 
15, no. 11, pp. 913?947, 2003. 
[40] S.-M. Park and J.-S. Yoo, ?Peer Reviewed: Electrochemical Impedance 
Spectroscopy for Better Electrochemical Measurements,? Anal. Chem., vol. 75, 
no. 21, pp. 455 A-461 A, Nov. 2003. 
[41] H. Ertu?rul and Z. O. Uygun, ?Impedimetric Biosensors for Label-Free and 
Enzymless Detection,? 2012, p. 17. 
[42] F. Lisdat and D. Sch?fer, ?The use of electrochemical impedance spectroscopy 
for biosensing,? Anal. Bioanal. Chem., vol. 391, no. 5, p. 1555, 2008. 
[43] ?Cystic Fibrosis Foundation Annual Patient Registry Report,? 2013. 
[44] P. D?te et al., ?Differentiating autoimmune pancreatitis from pancreatic 
cancer.,? Minerva Gastroenterol. Dietol., vol. 60, no. 4, pp. 247?253, 2014. 
[45] X. Kong, T. Sun, F. Kong, Y. Du, and Z. Li, ?Chronic Pancreatitis and 
Pancreatic Cancer,? Gastrointest. Tumors, vol. 1, no. 3, pp. 123?134, 2014. 
[46] R. Restrepo, H. E. Hagerott, S. Kulkarni, M. Yasrebi, and E. Y. Lee, ?Acute 
Pancreatitis in Pediatric Patients: Demographics, Etiology, and Diagnostic 
Imaging,? Am. J. Roentgenol., vol. 206, no. 3, pp. 632?644, 2016. 
156 
 
 
 
[47] K. Hanada et al., ?Effective screening for early diagnosis of pancreatic 
cancer,? Cancer Screen. Prev. Asia, vol. 29, no. 6, pp. 929?939, 2015. 
[48] P. D. Kamin, M. E. Bernardino, S. Wallace, and B.-S. Jing, ?Comparison of 
ultrasound and computed tomography in the detection of pancreatic 
malignancy,? Cancer, vol. 46, no. 11, pp. 2410?2412, Feb. 1980. 
[49] M. I. Canto et al., ?International Cancer of the Pancreas Screening (CAPS) 
Consortium summit on the management of patients with increased risk for 
familial pancreatic cancer,? Gut, vol. 62, no. 3, pp. 339?347, 2013. 
[50] T. Rustagi and P. A. Jamidar, ?Endoscopic Retrograde 
Cholangiopancreatography?Related Adverse Events: General Overview,? 
Minimizing, Recognizing, Manag. Endosc. Advers. Events, vol. 25, no. 1, pp. 
97?106, 2015. 
[51] M. Ozkan et al., ?Age-based computer-aided diagnosis approach for pancreatic 
cancer on endoscopic ultrasound images,? Endosc. Ultrasound, vol. 5, no. 2, 
pp. 101?107, 2016. 
[52] P. DK, B. HJ, J. Gyves, E. Allen, A. McLean, and P. DK, ?EValuation of a 
serologic marker, ca19-9, in the diagnosis of pancreatic cancer,? Ann. Intern. 
Med., vol. 110, no. 9, pp. 704?709, 1989. 
[53] A. Malesci et al., ?Determination of CA 19-9 antigen in serum and pancreatic 
juice for differential diagnosis of pancreatic adenocarcinoma from chronic 
pancreatitis,? Gastroenterology, vol. 92, no. 1, pp. 60?67, 1987. 
[54] M. Herreros-Villanueva and L. Bujanda, ?Non-invasive biomarkers in 
pancreatic cancer diagnosis: what we need versus what we have,? Ann. Transl. 
157 
 
 
 
Med., vol. 4, no. 7, p. 134, 2016. 
[55] H. Harada et al., ?Studies on Human Pure Pancreatic Juice Collected by 
Endoscopic Retrograde Catheterization of the Papilla,? Gastroentrologia Jpn., 
vol. 13, no. 5, pp. 383?389, 1978. 
[56] M. Tian et al., ?Proteomic analysis identifies MMP-9, DJ-1 and A1BG as 
overexpressed proteins in pancreatic juice from pancreatic ductal 
adenocarcinoma patients,? BMC Cancer, vol. 8, p. 241, 2008. 
[57] K. Matsumoto et al., ?Role of the preoperative usefulness of the pathological 
diagnosis of pancreatic diseases,? World J. Gastrointest. Oncol., vol. 8, no. 9, 
pp. 656?662, 2016. 
[58] S. Fleige and M. W. Pfaffl, ?RNA integrity and the effect on the real-time 
qRT-PCR performance,? Real-time Polym. Chain React., vol. 27, no. 2?3, pp. 
126?139, 2006. 
[59] L. Raeymaekers, ?Quantitative PCR: Theoretical Considerations with Practical 
Implications,? Anal. Biochem., vol. 214, no. 2, pp. 582?585, 1993. 
[60] S. Dastgheib, C. Irajie, R. Assaei, F. Koohpeima, and P. Mokarram, 
?Optimization of RNA Extraction from Rat Pancreatic Tissue,? Iran. J. Med. 
Sci., vol. 39, no. 3, pp. 282?288, 2014. 
[61] Z.-F. Quan, Z.-M. Wang, W.-Q. Li, and J.-S. Li, ?Use of endoscopic naso-
pancreatic drainage in the treatment of severe acute pancreatitis,? World J. 
Gastroenterol., vol. 9, no. 4, pp. 868?870, 2003. 
[62] G.-Y. Liou and P. Storz, ?Protein kinase D enzymes: novel kinase targets in 
pancreatic cancer,? Expert Rev. Gastroenterol. Hepatol., vol. 9, no. 9, pp. 
158 
 
 
 
1143?1146, 2015. 
[63] I. J. Zeitlin and W. Sircus, ?Factors influencing duodenal trypsin levels 
following a standard test meal as a test of pancreatic function,? Gut, vol. 15, 
no. 3, pp. 173?179, 1974. 
[64] M. E. Denyer and P. B. Cotton, ?Pure pancreatic juice studies in normal 
subjects and patients with chronic pancreatitis,? Gut, vol. 20, no. 2, pp. 89?97, 
1979. 
[65] R. Chen et al., ?Comparison of Pancreas Juice Proteins from Cancer Versus 
Pancreatitis Using Quantitative Proteomic Analysis,? Pancreas, vol. 34, no. 1, 
pp. 70?79, 2007. 
[66] J. V Olsen, S.-E. Ong, and M. Mann, ?Trypsin Cleaves Exclusively C-terminal 
to Arginine and Lysine Residues,? Mol. Cell. Proteomics, vol. 3, no. 6, pp. 
608?614, 2004. 
[67] J. Rodriguez, N. Gupta, R. D. Smith, and P. A. Pevzner, ?Does Trypsin Cut 
Before Proline?,? J. Proteome Res., vol. 7, no. 1, pp. 300?305, 2008. 
[68] H. U. Bermeyer, K. Gawehn, and M. Grassl, Methods of Enzymatic Analysis, 
2nd ed., vol. 1. New York, NY: Academic Press, Inc., 1974. 
[69] W. B. Corporation, Wothington Enzyme Manual, 1st ed. Lakewood, NJ, 1993. 
[70] C. Duan, M. A. Alibakhshi, D.-K. Kim, C. M. Brown, C. S. Craik, and A. 
Majumdar, ?Label-Free Electrical Detection of Enzymatic Reactions in 
Nanochannels,? ACS Nano, vol. 10, no. 8, pp. 7476?7484, 2016. 
[71] P. A. Neff, A. Serr, B. K. Wunderlich, and A. R. Bausch, ?Label-Free 
Electrical Determination of Trypsin Activity by a Silicon-on-Insulator Based 
159 
 
 
 
Thin Film Resistor,? ChemPhysChem, vol. 8, no. 14, pp. 2133?2137, 2007. 
[72] M. Stoytcheva, R. Zlatev, S. Cosnier, M. Arredondo, and B. Valdez, ?High 
sensitive trypsin activity evaluation applying a nanostructured QCM-sensor,? 
Biosens. Bioelectron., vol. 41, pp. 862?866, 2013. 
[73] L. Zhang, H. Qin, W. Cui, Y. Zhou, and J. Du, ?Label?free, turn?on 
fluorescent sensor for trypsin activity assay and inhibitor screening,? Talanta, 
vol. 161, pp. 535?540, 2016. 
[74] E. J. Herrera-L?pez, ?Lipase and Phospholipase Biosensors: A Review,? in 
Lipases and Phospholipases: Methods and Protocols, G. Sandoval, Ed. 
Totowa, NJ: Humana Press, 2012, pp. 525?543. 
[75] M. Kanimozhi, M. Johny, N. Gayathri, and R. Subashkumar, ?Optimization 
and Production of ? -Amylase from Halophilic Bacillus Species Isolated from 
Mangrove Soil Sources,? J. Appl. Environ. Microbiol., vol. 2, no. 3, pp. 70?73, 
2014. 
[76] M. J. Gibbs, A. Biela, and S. Krause, ??-Amylase sensor based on the 
degradation of oligosaccharide hydrogel films monitored with a quartz crystal 
sensor,? Spec. Issue Biosens. 2014, vol. 67, pp. 540?545, 2015. 
[77] J. Fortin, ?Transduction Principles,? in Functional Thin Films and 
Nanostructures for Sensors: Synthesis, Physics and Applications, A. Zribi and 
J. Fortin, Eds. Boston, MA: Springer US, 2009, pp. 17?29. 
[78] M. Stoytcheva, R. Zlatev, Z. Velkova, and G. Montero, ?Nanoparticle 
modified QCM-based sensor for lipase activity determination,? Anal. Methods, 
vol. 5, no. 16, pp. 3811?3815, 2013. 
160 
 
 
 
[79] N. F. Sheppard, M. J. Lesho, P. McNally, and A. Shaun Francomacaro, 
?Microfabricated conductimetric pH sensor,? Sensors Actuators B Chem., vol. 
28, no. 2, pp. 95?102, 1995. 
[80] ?Dielectric Behavior of Biomaterials at Different Frequencies on Room 
Temperature,? J. Phys. Conf. Ser., vol. 534, no. 1, p. 012063, 2014. 
[81] C. R. Mullins and C. A. Flood, ?A STUDY OF GASTRIC PEPSIN IN 
VARIOUS DISEASES.,? J. Clin. Invest., vol. 14, no. 6, pp. 793?7, Nov. 1935. 
[82] J. Mayerle, A. Stier, M. M. Lerch, and C.-D. Heidecke, ?Chronic Pancreatitis: 
Diagnostics and Therapy,? Surg., vol. 75, no. 7, pp. 731?748, Jul. 2004. 
[83] M. Kashiwagi et al., ?Group II and IV phospholipase A(2) are produced in 
human pancreatic cancer cells and influence prognosis.,? Gut, vol. 45, no. 4, 
pp. 605?12, Oct. 1999. 
[84] R. Rabanal, D. Fondevila, A. Vargas, A. Ramis, J. Badiola, and L. Ferrer, 
?Immunocytochemical detection of amylase, carboxypeptidase A, 
carcinoembryonic antigen and ?1-antitrypsin in carcinomas of the exocrine 
pancreas of the dog,? Res. Vet. Sci., vol. 52, no. 2, pp. 217?223, Mar. 1992. 
[85] I. L. Vovchuk and S. A. Petrov, ?The role of carboxypeptidases in 
carcinogenesis,? Biochem. Suppl. Ser. B Biomed. Chem., vol. 2, no. 3, pp. 267?
274, Sep. 2008. 
[86] J. Hermon-Taylor, J. Perrin, D. A. Grant, A. Appleyard, M. Bubel, and A. I. 
Magee, ?Immunofluorescent localisation of enterokinase in human small 
intestine.,? Gut, vol. 18, no. 4, pp. 259?65, Apr. 1977. 
[87] E. Lebenthal, I. Antonowicz, and H. Shwachman, ?Enterokinase and Trypsin 
161 
 
 
 
Activities in Pancreatic Insufficiency and Diseases of the Small Intestine,? 
Gastroenterology, vol. 70, no. 4, pp. 508?512, Apr. 1976. 
[88] T. ?uligoj, P. J. Ciclitira, and B. Bo?i?, ?Diagnostic and Research Aspects of 
Small Intestinal Disaccharidases in Coeliac Disease.,? J. Immunol. Res., vol. 
2017, p. 1042606, 2017. 
[89] ?Lactose intolerance - Symptoms and causes - Mayo Clinic.? [Online]. 
Available: https://www.mayoclinic.org/diseases-conditions/lactose-
intolerance/symptoms-causes/syc-20374232. [Accessed: 10-Apr-2019]. 
[90] S. R. Owens and J. K. Greenson, ?The pathology of malabsorption: current 
concepts,? Histopathology, vol. 50, no. 1, pp. 64?82, Jan. 2007. 
[91] R. Santini, J. Aviles, and T. W. Sheehy, ?Sucrase activity in the intestinal 
mucosa of patients with sprue and normal subjects,? Am. J. Dig. Dis., vol. 5, 
no. 12, pp. 1059?1062, Dec. 1960. 
[92] R. P. Verma and C. Hansch, ?Matrix metalloproteinases (MMPs): Chemical?
biological functions and (Q)SARs,? Bioorg. Med. Chem., vol. 15, no. 6, pp. 
2223?2268, Mar. 2007. 
[93] A. M. Knapinska, C.-A. Estrada, and G. B. Fields, ?The Roles of Matrix 
Metalloproteinases in Pancreatic Cancer,? Prog. Mol. Biol. Transl. Sci., vol. 
148, pp. 339?354, Jan. 2017. 
[94] H. Huang, Huang, and Hao, ?Matrix Metalloproteinase-9 (MMP-9) as a Cancer 
Biomarker and MMP-9 Biosensors: Recent Advances,? Sensors, vol. 18, no. 
10, p. 3249, Sep. 2018. 
[95] A. Biela et al., ?Disposable MMP-9 sensor based on the degradation of peptide 
162 
 
 
 
cross-linked hydrogel films using electrochemical impedance spectroscopy,? 
Biosens. Bioelectron., vol. 68, pp. 660?667, Jun. 2015. 
[96] C. Chen and X. Yang, ?The Diagnostic Effect of Capsule Endoscopy in 31 
Cases with Subacute Small Bowel Obstruction,? Gastrointest. Endosc., vol. 69, 
no. 5, p. AB216, Apr. 2009. 
[97] D. Friedel, R. Modayil, and S. Stavropoulos, ?Colon Capsule Endoscopy: 
Review and Perspectives,? Gastroenterol. Res. Pract., vol. 2016, p. 9643162, 
2016. 
[98] K. Kalantar-Zadeh et al., ?A human pilot trial of ingestible electronic capsules 
capable of sensing different gases in the gut,? Nat. Electron., vol. 1, no. 1, pp. 
79?87, Jan. 2018. 
[99] A. Nemiroski, M. Ryou, C. C. Thompson, and R. M. Westervelt, ?Swallowable 
fluorometric capsule for wireless triage of gastrointestinal bleeding,? Lab Chip, 
vol. 15, no. 23, pp. 4479?4487, 2015. 
[100] K. Kalantar-zadeh et al., ?Intestinal Gas Capsules: A Proof-of-Concept 
Demonstration,? Gastroenterology, vol. 150, no. 1, pp. 37?39, 2015. 
[101] N. OPTIZ and D. W. L?BBERS, ?Theory and Development of Fluorescence-
Based Optochemical Oxygen Sensors: Oxygen Optodes.,? Int. Anesthesiol. 
Clin., vol. 25, no. 3, 1987. 
[102] M. Mimee et al., ?An ingestible bacterial-electronic system to monitor 
gastrointestinal health.,? Science, vol. 360, no. 6391, pp. 915?918, May 2018. 
[103] C. M. Caffrey, K. Twomey, and V. I. Ogurtsov, ?Development of a wireless 
swallowable capsule with potentiostatic electrochemical sensor for 
163 
 
 
 
gastrointestinal track investigation,? Sensors Actuators B Chem., vol. 218, pp. 
8?15, 2015. 
[104] M. S. Arefin, J. M. Redoute, and M. R. Yuce, ?Integration of Low-Power 
ASIC and MEMS Sensors for Monitoring Gastrointestinal Tract Using a 
Wireless Capsule System,? IEEE J. Biomed. Heal. Informatics, vol. 22, no. 1, 
pp. 87?97, 2018. 
[105] C. R?mmele, J. Brueckner, H. Messmann, and S. G?lder, ?Clinical Experience 
with the PillCam Patency Capsule prior to Video Capsule Endoscopy: A Real-
World Experience,? Gastroenterol. Res. Pract., vol. 2016, pp. 1?6, Jan. 2016. 
[106] J. M. Park et al., ?Low-Power, High Data-Rate Digital Capsule Endoscopy 
Using Human Body Communication,? Applied Sciences , vol. 8, no. 9. 2018. 
[107] A. Kiourti, K. A. Psathas, and K. S. Nikita, ?Implantable and ingestible 
medical devices with wireless telemetry functionalities: A review of current 
status and challenges,? Bioelectromagnetics, vol. 35, no. 1, pp. 1?15. 
[108] A. Kiourti and K. S. Nikita, ?A Review of In-Body Biotelemetry Devices: 
Implantables, Ingestibles, and Injectables,? IEEE Trans. Biomed. Eng., vol. 64, 
no. 7, pp. 1422?1430, 2017. 
[109] V. Sayeed, ?Size, Shape, and Other Physical Attributes of Generic Tablets and 
Capsules Guidance for Industry,? 2015. 
[110] N.-K. Baek, I.-H. Sung, and D.-E. Kim, ?Frictional resistance characteristics of 
a capsule inside the intestine for microendoscope design,? Proc. Inst. Mech. 
Eng. Part H J. Eng. Med., vol. 218, no. 3, pp. 193?201, Mar. 2004. 
[111] ?Engineering surfaces to enhance cell adhesion: Surface engineering,? Mater. 
164 
 
 
 
Today, vol. 6, no. 11, p. 19, Nov. 2003. 
[112] H. W. Lu, Q. H. Lu, W. T. Chen, H. J. Xu, and J. Yin, ?Cell culturing on 
nanogrooved polystyrene petri dish induced by ultraviolet laser irradiation,? 
Mater. Lett., vol. 58, no. 1?2, pp. 29?32, Jan. 2004. 
[113] E. Biazar, M. Heidari, A. Asefnejad, A. Asefnezhad, and N. Montazeri, ?The 
relationship between cellular adhesion and surface roughness in polystyrene 
modified by microwave plasma radiation.,? Int. J. Nanomedicine, vol. 6, pp. 
631?9, 2011. 
[114] P. Ambe, S. A. Weber, M. Schauer, and W. T. Knoefel, ?Swallowed foreign 
bodies in adults.,? Dtsch. Arztebl. Int., vol. 109, no. 50, pp. 869?75, Dec. 2012. 
[115] CDRH, ?Use of International Standard ISO 10993-1, ?Biological evaluation of 
medical devices-Part 1: Evaluation and testing within a risk management 
process?; Guidance for Industry and Food and Drug Administration Staff 
Preface Public Comment,? 2016. 
[116] E. Trowers and M. E. Tischler, Gastrointestinal physiology?: a clinical 
approach. . 
[117] P. R. Slawinski, D. Oleynikov, and B. S. Terry, ?Intestinal biomechanics 
simulator for robotic capsule endoscope validation,? J. Med. Eng. Technol., 
vol. 39, no. 1, pp. 54?59, Jan. 2015. 
[118] N. K. Ahluwalia, D. G. Thompson, J. Barlow, and L. Heggie, ?Human small 
intestinal contractions and aboral traction forces during fasting and after 
feeding.,? Gut, vol. 35, no. 5, pp. 625?30, May 1994. 
[119] C. Roman, Ed., Gastrointestinal Motility. Dordrecht: Springer Netherlands, 
165 
 
 
 
1983. 
[120] B. Laulicht, A. Tripathi, V. Schlageter, P. Kucera, and E. Mathiowitz, 
?Understanding gastric forces calculated from high-resolution pill tracking.,? 
Proc. Natl. Acad. Sci. U. S. A., vol. 107, no. 18, pp. 8201?6, May 2010. 
[121] M. S. Arefin, J.-M. Redoute, and M. R. Yuce, ?Integration of Low-Power 
ASIC and MEMS Sensors for Monitoring Gastrointestinal Tract Using a 
Wireless Capsule System,? IEEE J. Biomed. Heal. Informatics, vol. 22, no. 1, 
pp. 87?97, Jan. 2016. 
[122] S. P. Woods and T. G. Constandinou, ?Wireless Capsule Endoscope for 
Targeted Drug Delivery: Mechanics and Design Considerations,? IEEE Trans. 
Biomed. Eng., vol. 60, no. 4, pp. 945?953, Apr. 2013. 
[123] A. D. Farmer, S. M. Scott, and A. R. Hobson, ?Gastrointestinal motility 
revisited: The wireless motility capsule.,? United Eur. Gastroenterol. J., vol. 1, 
no. 6, pp. 413?21, Dec. 2013. 
[124] G. Bhattacharya, A. Mathur, S. Pal, J. McLaughlin, and S. Roy, ?Equivalent 
Circuit Models and Analysis of Electrochemical Impedance Spectra of 
Caffeine Solutions and Beverages,? Int. J. Electrochem. Sci., vol. 11, pp. 
6370?6386, Jun. 2016. 
[125] M. Raimondo, M. Imoto, and E. P. Dimagno, ?Rapid endoscopic secretin 
stimulation test and discrimination of chronic pancreatitis and pancreatic 
cancer from disease controls,? Clin. Gastroenterol. Hepatol., vol. 1, no. 5, pp. 
397?403, 2003. 
[126] O. James, ?The Lundh test.,? Gut, vol. 14, no. 7, pp. 582?591, 1973. 
166 
 
 
 
[127] S. Pina, J. M. Oliveira, and R. L. Reis, ?Natural-Based Nanocomposites for 
Bone Tissue Engineering and Regenerative Medicine: A Review,? Adv. 
Mater., vol. 27, no. 7, pp. 1143?1169, Feb. 2015. 
[128] C.-H. Chuang et al., ?Detecting trypsin at liquid crystal/aqueous interface by 
using surface-immobilized bovine serum albumin,? Biosens. Bioelectron., vol. 
78, pp. 213?220, 2016. 
[129] C. Duan, M. A. Alibakhshi, D.-K. Kim, C. M. Brown, C. S. Craik, and A. 
Majumdar, ?Label-Free Electrical Detection of Enzymatic Reactions in 
Nanochannels,? ACS Nano, vol. 10, no. 8, pp. 7476?7484, 2016. 
[130] M. Stoytcheva, R. Zlatev, S. Cosnier, M. Arredondo, and B. Valdez, ?High 
sensitive trypsin activity evaluation applying a nanostructured QCM-sensor,? 
Biosens. Bioelectron., vol. 41, no. 1, pp. 862?866, 2013. 
[131] L.-Q. Wu, W. E. Bentley, and G. F. Payne, ?Biofabrication with biopolymers 
and enzymes: potential for constructing scaffolds from soft matter.,? Int. J. 
Artif. Organs, vol. 34, no. 2, pp. 215?24, Feb. 2011. 
[132] F. Teixeira Silva, B. Sorli, V. Calado, C. Guillaume, and N. Gontard, 
?Feasibility of a Gelatin Temperature Sensor Based on Electrical 
Capacitance.,? Sensors (Basel)., vol. 16, no. 12, Dec. 2016. 
[133] J. E. Eastoe, ?The amino acid composition of mammalian collagen and 
gelatin,? Biochem. J., vol. 61, no. 4, p. 589, 1955. 
[134] M. Li, ?Thin films of stimuli-responsive hydrogels.,? Polymers (Basel)., 2015. 
[135] Y. Yeo, W. Geng, T. Ito, D. S. Kohane, J. A. Burdick, and M. Radisic, 
?Photocrosslinkable hydrogel for myocyte cell culture and injection,? J. 
167 
 
 
 
Biomed. Mater. Res. Part B Appl. Biomater., vol. 81B, no. 2, pp. 312?322, 
2007. 
[136] B. Schyrr, S. Boder-Pasche, R. Ischer, R. Smajda, and G. Voirin, ?Fiber-optic 
protease sensor based on the degradation of thin gelatin films,? Sens. Bio-
Sensing Res., vol. 3, pp. 65?73, 2015. 
[137] A. Bigi, G. Cojazzi, S. Panzavolta, K. Rubini, and N. Roveri, ?Mechanical and 
thermal properties of gelatin films at different degrees of glutaraldehyde 
crosslinking,? Biomaterials, vol. 22, no. 8, pp. 763?768, 2001. 
[138] A. A. Justiz Vaillant and A. Jan, Physiology, Immune Response. 2019. 
[139] J. Li and D. J. Mooney, ?Designing hydrogels for controlled drug delivery,? 
Nat. Rev. Mater., vol. 1, no. 12, p. 16071, Dec. 2016. 
[140] F. Zang, K. Gerasopoulos, X. Z. Fan, A. D. Brown, J. N. Culver, and R. 
Ghodssi, ?Real-time monitoring of macromolecular biosensing probe self-
assembly and on-chip ELISA using impedimetric microsensors,? Biosens. 
Bioelectron., vol. 81, pp. 401?407, 2016. 
[141] Y. W. Kim et al., ?A surface acoustic wave biofilm sensor integrated with a 
treatment method based on the bioelectric effect,? Sensors Actuators A Phys., 
vol. 238, pp. 140?149, 2016. 
[142] Y. W. Kim et al., ?An ALD aluminum oxide passivated Surface Acoustic 
Wave sensor for early biofilm detection,? Sensors Actuators B Chem., vol. 163, 
no. 1, pp. 136?145, 2012. 
[143] S. Subramanian, ?Integrated Threshold-Activated Feedback Microsystem for 
Real-Time Characterization, Sensing, and Treatment of Bacterial Biofilms.,? 
168 
 
 
 
University of Maryland, College Park., 2016. 
[144] G. Banis, A. L. Beardslee, and R. Ghodssi, ?Gelatin-Enabled Microsensor for 
Pancreatic Trypsin Sensing,? Appl. Sci., vol. 8, no. 2, 2018. 
[145] N. Jaiswal, O. Prakash, M. Talat, S. H. Hasan, and R. K. Pandey, ??-Amylase 
immobilization on gelatin: Optimization of process variables,? J. Genet. Eng. 
Biotechnol., vol. 10, no. 1, pp. 161?167, 2012. 
[146] E. De Alteriis, P. Parascandola, S. Salvadore, and V. Scardi, ?Enzyme 
immobilisation within insolubilised gelatin,? J. Chem. Technol. Biotechnol. 
Biotechnol., vol. 35, no. 1, pp. 60?64, 1985. 
[147] N. W. Fadnavis, G. Sheelu, B. M. Kumar, M. U. Bhalerao, and A. A. 
Deshpande, ?Gelatin Blends with Alginate: Gels for Lipase Immobilization 
and Purification,? Biotechnol. Prog., vol. 19, no. 2, pp. 557?564, 2003. 
[148] N. Jaiswal and O. Prakash, ?Immobilization of Soybean ?-amylase on Gelatin 
and its Application as a Detergent Additive.,? Asian J. Biochem., vol. 6, pp. 
337?346, 2011. 
[149] M. L. Caldwell, E. S. Dickey, V. M. Hanrahan, H. C. Kung, J. T. Kung, and M. 
Misko, ?Amino Acid Composition of Crystalline Pancreatic Amylase from 
Swine1,? J. Am. Chem. Soc., vol. 76, no. 1, pp. 143?147, 1954. 
[150] J. D. BIANCHETTA, J. BIDAUD, A. A. GUIDONI, J. J. BONICEL, and M. 
ROVERY, ?Porcine Pancreatic Lipase,? Eur. J. Biochem., vol. 97, no. 2, pp. 
395?405, 1979. 
[151] K. V Thimann, ?THE EFFECT OF SALTS ON THE IONISATION OF 
GELATIN,? J. Gen. Physiol., vol. 14, no. 2, pp. 215?222, 1930. 
169 
 
 
 
[152] E. Barsoukov and R. Macdonald, Impedance Spectroscopy: Theory, 
Experiment, and Applications, 2nd ed. Wiley, 2005. 
[153] J. S. Daniels and N. Pourmand, ?Label-Free Impedance Biosensors: 
Opportunities and Challenges,? Electroanalysis, vol. 19, no. 12, pp. 1239?
1257, 2007. 
[154] Y. Erwanto et al., ?Microbial Transglutaminase Modifies Gel Properties of 
Porcine Collagen,? Asian-Australas J Anim Sci, vol. 16, no. 2, pp. 269?276, 
2003. 
[155] K. Franson and S. R?ssner, ?Fat intake and food choices during weight 
reduction with diet, behavioural modification and a lipase inhibitor,? J. Intern. 
Med., vol. 247, no. 5, pp. 607?614, 2000. 
[156] V. Konig, L. Vertesy, and T. R. Schneider, ?Structure of the alpha-amylase 
inhibitor tendamistat at 0.93 A,? Acta Crystallogr., vol. D, no. 59, pp. 1737?
1743, 2003. 
[157] C. Berggren, B. Bjarnason, and G. Johansson, ?Capacitive Biosensors,? 
Electroanalysis, vol. 13, no. 3, pp. 173?180, 2001. 
[158] T. L. Riss et al., Cell Viability Assays. Eli Lilly & Company and the National 
Center for Advancing Translational Sciences, 2004. 
[159] J. R. Crowther, The ELISA guidebook. Humana Press, 2010. 
[160] H. B. Brooks et al., Basics of Enzymatic Assays for HTS. Eli Lilly & Company 
and the National Center for Advancing Translational Sciences, 2004. 
[161] K. L. Cox, V. Devanarayan, A. Kriauciunas, J. Manetta, C. Montrose, and S. 
Sittampalam, Immunoassay Methods. Eli Lilly & Company and the National 
170 
 
 
 
Center for Advancing Translational Sciences, 2004. 
[162] E. K. Ulleberg, I. Comi, H. Holm, E. B. Herud, M. Jacobsen, and G. E. 
Vegarud, ?Human Gastrointestinal Juices Intended for Use in In Vitro 
Digestion Models.? 
[163] T. Ghosh, D. I. Lewis, A. T. R. Axon, and S. M. Everett, ?Review article: 
methods of measuring gastric acid secretion,? Aliment. Pharmacol. Ther., vol. 
33, no. 7, pp. 768?781, Apr. 2011. 
[164] W. H. Crosby, ?Small intestinal studies: Methods for obtaining intraluminal 
contents and intestinal mucosa,? Am. J. Dig. Dis., vol. 8, no. 1, pp. 2?11, Jan. 
1963. 
[165] Y. Amoako-Tuffour, M. L. Jones, N. Shalabi, A. Labbe, S. Vengallatore, and 
S. Prakash, ?Ingestible Gastrointestinal Sampling Devices: State-of-the-Art 
and Future Directions,? Crit. Rev. Biomed. Eng., vol. 42, no. 1, pp. 1?15, 2014. 
[166] I. Skala and J. Simacek, ?Device for obtaining samples of intestinal contents 
for microbiological and biochemical examinations,? Gut, vol. 9, pp. 246?248, 
1968. 
[167] G. E. Banis, L. A. Beardslee, J. M. Stine, and R. Ghodssi, ?Enteric & 3D-
Printed Hybrid Package for Sampling in Digestive Regions,? in A Solid-State 
Sensors, Actuators and Microsystems Workshop, 2018, pp. 96?97. 
[168] J. L. Gonzalez-Guillaumin, D. C. Sadowski, K. V. I. S. Kaler, and M. P. 
Mintchev, ?Ingestible Capsule for Impedance and pH Monitoring in the 
Esophagus,? IEEE Trans. Biomed. Eng., vol. 54, no. 12, pp. 2231?2236, 2007. 
[169] P. Demosthenous and J. Georgiou, ?An ingestible, NIR-fluorometric, cancer-
171 
 
 
 
screening capsule,? in 2015 37th Annual International Conference of the IEEE 
Engineering in Medicine and Biology Society (EMBC), 2015, pp. 2143?2146. 
[170] W. Yu, R. Rahimi, M. Ochoa, R. Pinal, and B. Ziaie, ?A Smart Capsule With 
GI-Tract-Location-Specific Payload Release,? IEEE Trans. Biomed. Eng., vol. 
62, no. 9, pp. 2289?2295, 2015. 
[171] S. S. Mapara and V. B. Patravale, ?Medical capsule robots: A renaissance for 
diagnostics, drug delivery and surgical treatment,? J. Control. Release, vol. 
261, pp. 337?351, 2017. 
[172] C. M. Caffrey, O. Chevalerias, C. O?Mathuna, and K. Twomey, ?Swallowable-
Capsule Technology,? IEEE Pervasive Comput., vol. 7, no. 1, pp. 23?29, 2008. 
[173] K. Kalantar-zadeh, N. Ha, J. Z. Ou, and K. J. Berean, ?Ingestible Sensors,? 
ACS Sensors, vol. 2, no. 4, pp. 468?483, 2017. 
[174] S. N. Adler and Y. C. Metzger, ?PillCam COLON capsule endoscopy: recent 
advances and new insights,? Therap. Adv. Gastroenterol., vol. 4, no. 4, pp. 
265?268, 2011. 
[175] M. Rasouli et al., ?Therapeutic Capsule Endoscopy: Opportunities and 
Challenges,? J. Healthc. Eng., vol. 2, no. 4, 2011. 
[176] G. Pan and L. Wang, ?Swallowable Wireless Capsule Endoscopy: Progress 
and Technical Challenges,? Gastroenterol. Res. Pract., vol. 2012, pp. 1?9, 
2012. 
[177] C. Van de Bruaene, D. De Looze, and P. Hindryckx, ?Small bowel capsule 
endoscopy: Where are we after almost 15 years of use?,? World J. 
Gastrointest. Endosc., vol. 7, no. 1, pp. 13?36, Jan. 2015. 
172 
 
 
 
[178] J. Wors?e et al., ?Gastric transit and small intestinal transit time and motility 
assessed by a magnet tracking system,? BMC Gastroenterol., vol. 11, no. 1, p. 
145, Dec. 2011. 
[179] N. Mehmood and S. M. Aziz, ?Magnetic sensing technology for in vivo 
tracking,? in 2012 International Conference on Emerging Technologies, 2012, 
pp. 1?4. 
[180] I. Umay, B. Fidan, and B. Barshan, ?Localization and Tracking of Implantable 
Biomedical Sensors,? Sensors (Basel)., vol. 17, no. 3, 2017. 
[181] E. A. Johannessen, L. Wang, S. W. J. Reid, D. R. S. Cumming, and J. M. 
Cooper, ?Implementation of radiotelemetry in a lab-in-a-pill format,? Lab 
Chip, vol. 6, no. 1, pp. 39?45, Dec. 2006. 
[182] M. Yu, ?M2A capsule endoscopy. A breakthrough diagnostic tool for small 
intestine imaging.,? Gastroenterol. Nurs., vol. 25, no. 1, pp. 24?7. 
[183] Y. Ye et al., ?Bounds on Rf Cooperative Localization for Video Capsule 
Endoscopy.? 2013. 
[184] V. V. Khutoryanskiy, ?Longer and safer gastric residence,? Nat. Mater., vol. 
14, no. 10, pp. 963?964, Oct. 2015. 
[185] D. F. Evans, G. Pye, R. Bramley, A. G. Clark, T. J. Dyson, and J. D. 
Hardcastle, ?Measurement of gastrointestinal pH profiles in normal ambulant 
human subjects.,? Gut, vol. 29, no. 8, pp. 1035?41, Aug. 1988. 
[186] J. L. Seifter and H.-Y. Chang, ?Extracellular Acid-Base Balance and Ion 
Transport Between Body Fluid Compartments,? Physiology, vol. 32, no. 5, pp. 
367?379, Sep. 2017. 
173 
 
 
 
[187] J. R. Taylor, C. A. Cooper, and T. P. Mommsen, ?Implications of GI function 
for gas exchange, acid?base balance and nitrogen metabolism,? Fish Physiol., 
vol. 30, pp. 213?259, Jan. 2010. 
[188] S. G. NUGENT, D. KUMAR, D. S. RAMPTON, and D. F. EVANS, 
?Intestinal luminal pH in inflammatory bowel disease: possible determinants 
and implications for therapy with aminosalicylates and other drugs,? Gut, vol. 
48, no. 4, pp. 571?577, Apr. 2001. 
[189] D. Lalezari, ?Gastrointestinal pH profile in subjects with irritable bowel 
syndrome,? Ann. Gastroenterol., vol. 25, no. 4, p. 333, 2012. 
[190] T. P. Bennett, Modern topics in biochemistry?: structure and function of 
biological molecules. New York?: The Macmillan Company?;, 1969. 
[191] L. R. Marek, ?Elements of General and Biological Chemistry (Holum, John 
R.),? J. Chem. Educ., vol. 60, no. 7, p. 613, Jul. 1983. 
[192] M. Puccetti, S. Giovagnoli, T. Zelante, L. Romani, and M. Ricci, 
?Development of novel indole-3-aldehyde loaded gastro-resistant spray-dried 
microparticles for post-biotic small intestine local delivery,? J. Pharm. Sci., 
2018. 
[193] E. T. Cole et al., ?Enteric coated HPMC capsules designed to achieve intestinal 
targeting,? Int. J. Pharm., vol. 231, no. 1, pp. 83?95, 2002. 
[194] K. Whitehead, Z. Shen, and S. Mitragotri, ?Oral delivery of macromolecules 
using intestinal patches: applications for insulin delivery,? J. Control. Release, 
vol. 98, no. 1, pp. 37?45, 2004. 
[195] G. E. Banis, S. Member, L. A. Beardslee, J. M. Stine, R. M. Sathyam, and R. 
174 
 
 
 
Ghodssi, ?Gastrointestinal Targeted Sampling and Sensing via Embedded 
Packaging of Integrated Capsule System,? J. Microelectromechanical Syst., 
vol. PP, pp. 1?7, 2019. 
[196] L. P. Degen and S. F. Phillips, ?Variability of gastrointestinal transit in healthy 
women and men.,? Gut, vol. 39, no. 2, pp. 299?305, Aug. 1996. 
[197] L. R. Johnson, Physiology of the gastrointestinal tract. Academic Press, 2012. 
[198] M. Camilleri and C. M. Prather, ?Axial Forces During Gastric Emptying in 
Health and Models of Disease,? 1994. 
[199] D. M. Mudie, G. L. Amidon, and G. E. Amidon, ?Physiological Parameters for 
Oral Delivery and in Vitro Testing,? Mol. Pharm., vol. 7, no. 5, pp. 1388?
1405, Oct. 2010. 
[200] C. P. Gayer and M. D. Basson, ?The effects of mechanical forces on intestinal 
physiology and pathology.,? Cell. Signal., vol. 21, no. 8, pp. 1237?44, Aug. 
2009. 
[201] B. Laulicht, A. Tripathi, V. Schlageter, P. Kucera, and E. Mathiowitz, 
?Understanding gastric forces calculated from high-resolution pill tracking.,? 
Proc. Natl. Acad. Sci. U. S. A., vol. 107, no. 18, pp. 8201?6, May 2010. 
[202] K. J. Berean et al., ?Enhanced Gas Permeation through Graphene 
Nanocomposites,? J. Phys. Chem. C, vol. 119, no. 24, pp. 13700?13712, Jun. 
2015. 
[203] M. E. Haupt, M. J. Kwasny, M. S. Schechter, and S. A. McColley, ?Pancreatic 
Enzyme Replacement Therapy Dosing and Nutritional Outcomes in Children 
with Cystic Fibrosis,? J. Pediatr., vol. 164, no. 5, pp. 1110-1115.e1, 2014. 
175 
 
 
 
[204] R. Muench and R. Ammann, ?Fecal Immunoreactive Lipase: A New Tubeless 
Pancreatic Function Test,? Scand. J. Gastroenterol., vol. 27, no. 4, pp. 289?
294, Jan. 1992. 
[205] P. T. Regan and E. P. DiMagno, ?Exocrine pancreatic insufficiency in celiac 
sprue: A cause of treatment failure,? Gastroenterology, vol. 78, no. 3, pp. 484?
487, Mar. 1980. 
[206] C. Basnayake and D. Ratnam, ?Blood tests for acute pancreatitis.,? Aust. 
Prescr., vol. 38, no. 4, pp. 128?30, Aug. 2015. 
[207] J. Walkowiak et al., ?[Comparison of faecal lipase test and faecal elastase-1 
test in the assessment of exocrine pancreatic function in cystic fibrosis].,? Med. 
Wieku Rozwoj., vol. 7, no. 2, pp. 149?55. 
[208] C. Forsmark and P. C. Adams, ?Pancreatic function testing--valuable but 
underused.,? Can. J. Gastroenterol., vol. 23, no. 8, pp. 529?30, Aug. 2009. 
[209] P. A. Hart et al., ?Endoscopic Pancreas Fluid Collection: Methods and 
Relevance for Clinical Care and Translational Science.,? Am. J. Gastroenterol., 
vol. 111, no. 9, pp. 1258?66, 2016. 
[210] M. Vujasinovic, R. Valente, M. Del Chiaro, J. Permert, and J.-M. L?hr, 
?Pancreatic Exocrine Insufficiency in Pancreatic Cancer.,? Nutrients, vol. 9, 
no. 3, Feb. 2017. 
[211] J. E. Dom?nguez-Mu?oz, L. Nieto-Garcia, J. L?pez-D?az, J. Lari?o-Noia, I. 
Abdulkader, and J. Iglesias-Garcia, ?Impact of the treatment of pancreatic 
exocrine insufficiency on survival of patients with unresectable pancreatic 
cancer: a retrospective analysis.,? BMC Cancer, vol. 18, no. 1, p. 534, May 
176 
 
 
 
2018. 
[212] ?A Beginners Guide to Ion-Selective Electrodes. All you need to know about 
theory and practice of ISE measurements, with comprehensive Electro-
chemical Glossary.? [Online]. Available: 
http://www.nico2000.net/Book/Guide1.html. [Accessed: 22-Apr-2019]. 
[213] Z. Xiao, R. Lopez, M. A. Parsi, M. Dodig, and T. Stevens, ?Comparison of 
autoanalyzer and back titration for measurement of bicarbonate concentration 
in endoscopically collected pancreatic fluid.,? Pancreas, vol. 40, no. 2, pp. 
237?41, Mar. 2011. 
[214] E. J. Herrera-L?pez, ?Lipase and Phospholipase Biosensors: A Review,? 
Humana Press, 2012, pp. 525?543. 
[215] K. Ge, D. H. Liu, K. Chen, L. H. Nie, and S. Z. Yao, ?Assay of Pancreatic 
Lipase with the Surface Acoustic Wave Sensor System,? Anal. Biochem., vol. 
226, no. 2, pp. 207?211, Apr. 1995. 
[216] I. Ben Rejeb, F. Arduini, A. Amine, M. Gargouri, and G. Palleschi, 
?Amperometric biosensor based on Prussian Blue-modified screen-printed 
electrode for lipase activity and triacylglycerol determination,? Anal. Chim. 
Acta, vol. 594, no. 1, pp. 1?8, Jun. 2007. 
[217] M. S Bhutani, V. Gupta, S. Guha, D. Gheonea, and A. Saftoiu, Pancreatic cyst 
fluid analysis?A review, vol. 20. 2011. 
[218] P. Vijayvargiya, M. Camilleri, A. Shin, and A. Saenger, ?Methods for 
diagnosis of bile acid malabsorption in clinical practice.,? Clin. Gastroenterol. 
Hepatol., vol. 11, no. 10, pp. 1232?9, Oct. 2013. 
177 
 
 
 
[219] C. Degirolamo, S. Modica, G. Palasciano, and A. Moschetta, ?Bile acids and 
colon cancer: Solving the puzzle with nuclear receptors,? Trends Mol. Med., 
vol. 17, no. 10, pp. 564?572, Oct. 2011. 
[220] K. Cheng and J.-P. Raufman, ?Bile acid-induced proliferation of a human 
colon cancer cell line is mediated by transactivation of epidermal growth factor 
receptors,? Biochem. Pharmacol., vol. 70, no. 7, pp. 1035?1047, Oct. 2005. 
[221] L. WEDLAKE, R. A?HERN, D. RUSSELL, K. THOMAS, J. R. F. 
WALTERS, and H. J. N. ANDREYEV, ?Systematic review: the prevalence of 
idiopathic bile acid malabsorption as diagnosed by SeHCAT scanning in 
patients with diarrhoea-predominant irritable bowel syndrome,? Aliment. 
Pharmacol. Ther., vol. 30, no. 7, pp. 707?717, Oct. 2009. 
[222] G. Lake-Bakaar, S. McKavanagh, C. E. Rubio, O. Epstein, and J. A. 
Summerfield, ?Measurement of trypsin in duodenal juice by 
radioimmunoassay.,? Gut, vol. 21, no. 5, pp. 402?7, May 1980. 
[223] American Chemical Society., Reagent chemicals?: American Chemical Society 
specifications, official from April 1, 1993. The Society, 1993. 
[224] American Society of Biological Chemists., Rockefeller Institute for Medical 
Research., and American Society for Biochemistry and Molecular Biology., 
The Journal of biological chemistry. American Society for Biochemistry and 
Molecular Biology. 
[225] A. Robles Medina, L. Esteban Cerd?n, A. Gim?nez Gim?nez, B. Camacho 
P?ez, M. J. Ib??ez Gonz?lez, and E. Molina Grima, ?Lipase-catalyzed 
esterification of glycerol and polyunsaturated fatty acids from fish and 
178 
 
 
 
microalgae oils,? J. Biotechnol., vol. 70, no. 1?3, pp. 379?391, Apr. 1999. 
[226] P. Speranza, A. P. B. Ribeiro, and G. A. Macedo, ?Lipase catalyzed 
interesterification of Amazonian patau? oil and palm stearin for preparation of 
specific-structured oils.,? J. Food Sci. Technol., vol. 52, no. 12, pp. 8268?75, 
Dec. 2015. 
[227] Y. K. Lee et al., ?Kinetics and Chemistry of Hydrolysis of Ultrathin, 
Thermally Grown Layers of Silicon Oxide as Biofluid Barriers in Flexible 
Electronic Systems,? ACS Appl. Mater. Interfaces, vol. 9, no. 49, pp. 42633?
42638, Dec. 2017. 
[228] D. Bonn, ?Wetting transitions,? Curr. Opin. Colloid Interface Sci., vol. 6, no. 
1, pp. 22?27, Feb. 2001. 
[229] M.-C. Michalski and B. J. V. Saramago, ?Static and Dynamic Wetting 
Behavior of Triglycerides on Solid Surfaces,? J. Colloid Interface Sci., vol. 
227, no. 2, pp. 380?389, Jul. 2000. 
[230] L. D. A. Chumpitaz, L. F. Coutinho, and A. J. A. Meirelles, ?Surface tension 
of fatty acids and triglycerides,? J. Am. Oil Chem. Soc., vol. 76, no. 3, pp. 379?
382, Mar. 1999. 
[231] W. Xie et al., ?Design and Validation of a Biosensor Implantation Capsule 
Robot,? J. Biomech. Eng., vol. 139, no. 8, p. 081003, Jun. 2017. 
[232] J. Carr?re, C. Galabert, J. P. Thouvenot, and C. Figarella, ?Assay of human 
pancreatic lipase in biological fluids using a non-competitive enzyme 
immunoassay.,? Clin. Chim. Acta., vol. 161, no. 2, pp. 209?19, Dec. 1986. 
[233] K. E. Barrett and W. F. Ganong, Ganong?s review of medical physiology. LK  - 
179 
 
 
 
https://umaryland.on.worldcat.org/oclc/779244271, 24th ed. /. New York: 
McGraw-Hill Medical?;, 2012. 
[234] M. Camilleri et al., ?Human gastric emptying and colonic filling of solids 
characterized by a new method,? Am. J. Physiol. - Gastrointest. Liver Physiol., 
vol. 257, no. 2, 1989. 
[235] Z. Huang et al., ?Effects of triglycerides levels in human whole blood on the 
extraction of 19 commonly used drugs using liquid-liquid extraction and gas 
chromatography-mass spectrometry.,? Toxicol. reports, vol. 2, pp. 785?791, 
2015. 
[236] Y. K. Lee et al., ?Kinetics and Chemistry of Hydrolysis of Ultrathin, 
Thermally Grown Layers of Silicon Oxide as Biofluid Barriers in Flexible 
Electronic Systems,? ACS Appl. Mater. Interfaces, vol. 9, no. 49, pp. 42633?
42638, Dec. 2017. 
[237] A. Ziegler, L. Gonzalez, and A. Blikslager, ?Large Animal Models: The Key to 
Translational Discovery in Digestive Disease Research,? CMGH, vol. 2, no. 6. 
Elsevier Inc, pp. 716?724, 01-Nov-2016. 
[238] J. Flemming and S. Cameron, ?Small bowel capsule endoscopy: Indications, 
results, and clinical benefit in a University environment,? Medicine 
(Baltimore)., vol. 97, no. 14, pp. e0148?e0148, Apr. 2018. 
[239] P. R. Slawinski, D. Oleynikov, and B. S. Terry, ?Intestinal biomechanics 
simulator for robotic capsule endoscope validation,? J. Med. Eng. Technol., 
vol. 39, no. 1, pp. 54?59, Jan. 2015. 
[240] R. Miftahof and N. Akhmadeev, ?Dynamics of intestinal propulsion,? J. Theor. 
180 
 
 
 
Biol., vol. 246, no. 2, pp. 377?393, May 2007. 
[241] W. L. Hasler, ?The use of SmartPill for gastric monitoring,? Expert Rev. 
Gastroenterol. Hepatol., vol. 8, no. 6, pp. 587?600, Aug. 2014. 
[242] H. Y. Shi, S. C. Ng, K. K. Tsoi, J. C. Wu, J. J. Sung, and F. K. Chan, ?The role 
of capsule endoscopy in assessing mucosal inflammation in ulcerative colitis,? 
Expert Rev. Gastroenterol. Hepatol., vol. 9, no. 1, pp. 47?54, Jan. 2015. 
[243] S. H. Itzkowitz, D. H. Present, and Crohn?s and Colitis Foundation of America 
Colon Cancer in IBD Study Group, ?Consensus conference: Colorectal cancer 
screening and surveillance in inflammatory bowel disease.,? Inflamm. Bowel 
Dis., vol. 11, no. 3, pp. 314?21, Mar. 2005. 
[244] J. K. Limdi and F. A. Farraye, ?An Update on Surveillance in Ulcerative 
Colitis.? 
[245] Y. L. Kong et al., ?3D?Printed Gastric Resident Electronics,? Adv. Mater. 
Technol., vol. 4, no. 3, p. 1800490, Dec. 2018. 
[246] P. G. Toner, K. E. Carr, A. Ferguson, and C. Mackay, ?Scanning and 
transmission electron microscopic studies of human intestinal mucosa.,? Gut, 
vol. 11, no. 6, pp. 471?81, Jun. 1970. 
[247] T. Skrzypek et al., ?Light and scanning electron microscopy evaluation of the 
postnatal small intestinal mucosa development in pigs.,? J. Physiol. 
Pharmacol., vol. 56 Suppl 3, pp. 71?87, Jun. 2005. 
[248] A. Menciassi and P. Dario, ?Bio-inspired solutions for locomotion in the 
gastrointestinal tract: background and perspectives,? Philos. Trans. R. Soc. 
London. Ser. A Math. Phys. Eng. Sci., vol. 361, no. 1811, pp. 2287?2298, Oct. 
181 
 
 
 
2003. 
[249] W. Xie et al., ?Design and Validation of a Biosensor Implantation Capsule 
Robot,? J. Biomech. Eng., vol. 139, pp. 081003?1, 2017. 
[250] M. D. Breton et al., ?Continuous Glucose Monitoring and Insulin Informed 
Advisory System with Automated Titration and Dosing of Insulin Reduces 
Glucose Variability in Type 1 Diabetes Mellitus,? Diabetes Technol. Ther., 
vol. 20, no. 8, pp. 531?540, Aug. 2018. 
[251] C. Steiger, A. Abramson, P. Nadeau, A. P. Chandrakasan, R. Langer, and G. 
Traverso, ?Ingestible electronics for diagnostics and therapy,? Nat. Rev. 
Mater., vol. 4, no. 2, pp. 83?98, Feb. 2019. 
[252] I. Bilican, M. T. Guler, N. Gulener, M. Yuksel, and S. Agan, ?Capacitive 
solvent sensing with interdigitated microelectrodes,? Microsyst. Technol., vol. 
22, no. 3, pp. 659?668, 2016. 
 
182