UnE

n RD
2O19
AUG 13

UNDERGRADUATE
ENGINEERING
RESEARCH DAY
 Thank you to our 2019 sponsors:

01 • UnERD
2O19
AUG 13
Message From
the Vice Dean
June 5, 2019

Engineering students have shared their summer

research through poster and podium presentations
during the Undergraduate Engineering Research Day
(UnERD) for more than a decade. The Faculty of

4
Applied Science and Engineering endorses this
unique opportunity to celebrate and promote under-
graduate research.

The Faculty is extremely proud of this student-led initiative, which

provides our undergraduate students the opportunity to experience the
excitement of participating in a research conference, presenting their

6
results before their peers, a panel of judges consisting of our alumni and
faculty, as well as industry colleagues. The process of developing oral
presentations and posters prepares participants not only for future aca-
demic conferences, but for a range of opportunities to present their work
and themselves to diverse audiences. Many of our winners have gone on
to excel in their academic and professional pursuits, and we expect the
same will be true for this year’s participants.

7
UnERD is an event that acknowledges students’ hard work, while provid-
ing an opportunity for new and well-established companies to network
with bright young minds. Your participation and support motivate
students to strive for excellence and encourages the future generation of
researchers.

9
Thomas W. Coyle
Sc. D. Vice Dean, Undergraduate
UnERD
2O19 • 02
AUG 13
 Bahen 1st Floor

BA1190 BA1180 BA1170 BA1160 BA1130

Entrance

St George St
BA BA BA
1220 1210 1200

Registration Table
Podium Presentation Room
TABLE OF CONTENTS
MESSAGE FROM THE ORGANIZERS 5
AGENDA 6
KEYNOTE SPEAKERS 7
ABSTRACTS - PODIUM PRESENTATIONS
A: CHEMICAL ...............................................................................................................09
B: BIOMATERIALS & BIOMEDICAL ....................................................................14
8
C: BIOENGINEERING.................................................................................................18
D: MATERIALS SCIENCE ........................................................................................20
E: ELECTRICAL ...........................................................................................................28
F: MECHANICAL & INDUSTRIAL ........................................................................31
G: AEROSPACE ...........................................................................................................36
H: ENGINEERING EDUCATION ............................................................................37

ABSTRACTS - POSTER PRESENTATIONS

A: CHEMICAL ..............................................................................................................39
B: BIOMATERIALS & BIOMEDICAL ...................................................................42
38
C: BIOENGINEERING & BIOELECTRICAL ......................................................54
D: MATERIALS SCIENCE .......................................................................................57
E: ELECTRICAL & COMPUTER ...........................................................................61
F: MECHANICAL & INDUSTRIAL .......................................................................65
G: CIVIL/MINERAL ...................................................................................................72
H: FOOD AND NUTRITION .....................................................................................74

ACKNOWLEGEMENTS ACKNOWLEDGEMENTS
76
INNOVATION
BEGINS
WITH
YOU
WELCOME to the University of Toronto’s Undergraduate Engineering Research Day, UnERD 2019! UnERD is
the annual student-run research conference that provides young aspiring researchers the opportunity to
showcase their summer research accomplishments to industry professionals and fellow students. UnERD
2019 celebrates undergraduate engineering research within the Faculty of Applied Science & Engineering
at the University of Toronto, with a specific focus on the research projects that inspire the exchange of
innovative and impactful solutions across a variety of global challenges.

Throughout the day, ___

45 poster and 37
__ podium presentations from a diverse range of engineering disciplines
will be evaluated by professors and graduate students from the Faculty of Applied Science & Engineering.
We hope the undergraduate researchers will gain conference experience in presenting their research
projects and establishing connections with experts in their respective fields. This year’s addition of an
interactive exhibit by the research group, BME-In-Your-Hands, will allow students to learn about the
human body through physiological sensors, providing an open source learning platform for engineering
students.

We would like to welcome Professor Christopher Yip, Dean of the Faculty of Applied Science & Engineering,
as our opening keynote speaker, and Professor Jason Anderson as our closing keynote speaker. We are
also pleased to welcome Professor Thomas Coyle, the Vice Dean of Undergraduate Studies for the Faculty
of Applied Science & Engineering, to give closing remarks on behalf of the Faculty, which graciously funds
and supports UnERD.

Our gratitude goes to our sponsors: The Faculty of Applied Science & Engineering; The Department of
Electrical & Computer Engineering; The Department of Mechanical & Industrial Engineering; and Galbraith
Society. We would also like to extend our thanks to all the students for participating in the conference, the
judges for evaluating the presentations and exchanging meaningful concepts with the students, and the
volunteers for making this day possible!

Thank you for attending UnERD 2019, and we hope you enjoy the conference!

Sincerely,
UnERD 2019 Planning Committee

05 • UnERD
2O19
AUG 13
 Registration / Bahen Atrium
8:00 am - 9:00 am
Registration & Breakfast

Opening Ceremony / BA1130

9:00 am - 9: 10 am
Co-Chair Presentation
9:10 am -9:20 am
Sponsor Presentation
9:20 am - 9:35 am
Keynote Speaker: Dean Christopher Yip
9:35 am - 9:45 am
Transition

UNDERGRADUATE Podium Presentations & Interactive Exibit

ENGINEERING RESEARCH DAY 9:45 am - 11:55 am

AGENDA
Podium Presentations & Interactive Exibit

Lunch / Bahen Atrium

11:55 am - 12:50 pm
BAHEN CENTRE Lunch + Networking
AUGUST 13, 2019 12:50 pm - 1:00 pm
Transition

Poster Presentations / Bahen Atrium

1:00 pm - 2:50 pm
Presentations
2:50 pm - 3:00 pm
Transition

Podium Presentations
3:00 pm - 4:50 pm
Presentations
4:50 pm - 5:00 pm
Transition

Closing Ceremony / BA1130

5:00 pm - 5:05 pm
Co-Chair Presentation
5:05 pm - 5:20 pm
Keynote Speaker: Professor Jason Anderson
5:20 pm - 5:35 pm
Closing Remarks: Professor Thomas Coyle
5:35 pm - 5:50 pm
Awards
5:50 pm - 6:00 pm
Acknowledgements

UnERD
2O19 • 06
AUG 13
 Opening Keynote Speaker Time: 9:35 am
Location: BA 1130
Dean Christopher Yip
Christopher Yip is the new Dean of the Faculty of Applied Science &
Engineering at the University of Toronto and a professor in the Depart-
ment of Chemical Engineering and Biomedical Engineering, with
cross-appointments to Biochemistry and the Donnelly Centre for Cellu-
lar and Biomolecular Research.

Dean Yip previously served as the University’s Associate Vice-President

of International Partnership and the Director of the Institute of Biomate-
rials and Biomedical Engineering. He is an internationally renowned
scholar for his research work on characterization of biomolecular
processes and structures by applying molecular imaging techniques. He
received numerous awards including Premier’s Research Excellence in
1999, University of Toronto Faculty of Engineering and Applied Science
Early Career Faculty Teaching Award in 2000, and held Tier II Canada
Research Chair Award in molecular imaging from 2000 -2010.

Closing Keynote Speaker Time: 5:35 pm

Location: BA 1130
Professor Jason Anderson
Jason Anderson is the Associate Chair (Research) and a professor in the
Department of Electrical and Computer Engineering at the University of
Toronto. He is Director of the minor and certificate programs in Artificial
Intelligence Engineering and is a Faculty Affiliate of the Vector Institute
for Artificial Intelligence.

As the Principal Investigator of Programmable Digital Research Group,

Professor Anderson’s research focuses on programmable hardware
design and aims to optimize speed, energy-consumption and
ease-of-use. In 2014, his research group received the Community Award
at FPL 2014 for “contributions to open-source high-level synthesis” and
he received Ontario Early Researcher Award from the Ministry of
Research and Innovation. Professor Anderson also received the Universi-
ty of Toronto Faculty of Engineering and Applied Science Early Career
Faculty Teaching Award in 2013.

07 • UnERD
2O19
AUG 13
CONTENTS: ABSTRACTS - PODIUM PRESENTATIONS
A: CHEMICAL .....................................................................................09
B: BIOMATERIALS & BIOMEDICAL ..............................................14
C: BIOENGINEERING.........................................................................18
D: MATERIALS SCIENCE ..................................................................20
E: ELECTRICAL ..................................................................................28
F: MECHANICAL & INDUSTRIAL ...................................................31
G: AEROSPACE ...................................................................................36
H: ENGINEERING EDUCATION .......................................................37
Abstracts – Podium Presentations
A: Chemical
A1: Extraction of rare earth elements from Canadian ore using supercritical fluid extraction
Kimberly Watada1 , Jiakai Zhang1 , Gisele Azimi1,2

Laboratory for Strategic Materials, Department of Chemical Engineering and Applied Chemistry, University of Toronto;
1

Department of Materials Science and Engineering, University of Toronto

2

Rare Earth Elements (REEs) are crucial raw materials that are widely used in many consumer electronics and renew-
able energy technologies. Over recent years, the global demandfor REEs has increased which represents an opportu-
nity for Canada to enter the emerging market. Conventionally, REEs are extracted from Canadian ores using hydro-
metallurgy or pyrometallurgy. These methods face environmental challenges, as the former produces hazardous
acidic waste, and the latter requires high amounts of energy and emits greenhouse gases. Therefore, an environmen-
tally and economically efficient extraction method for REEs from primary sources has become increasingly
parmount. Supercritical fluid extraction (SCFE) has emerged as a green alternative for the extraction of REEs dueto
the favourable properties of supercritical fluids as solvents. In particular, supercritical carbon dioxide (sc-CO2) is
inert, abundant, cost effective and easily recyclable [1] . This study uses sc-CO2 to extract REEs complexed with
tributyl-phosphate-nitric acid from a Canadian ore and evaluate the extraction efficiency of the process. Utilizing
fractional factorial design, the effect of temperature, solid to chelating agent ratio, pressure, time, and agitation on
the percent extraction efficiency arestudied. With the results, an empirical model for the extraction was constructed
with linear least squares regression. Optimized process parameter selection will be calculated based on the model
and verified with validation tests. Approximately 60% and 75% REE extraction rates were achieved from ore and
pretreated ore samples respectively. The results of this study will provide information on the feasibility of SCFE as
a sustainable alternative for REE extraction from primary sources.

[1] J. Zhang, J. Anawati, Y. Yao, and G. Azimi, “Aeriometallurgical Extraction of Rare Earth Elements from a
NdFeB Magnet Utilizing Supercritical Fluids,” ACS Sustainable Chemistry & Engineering, 2018, 6 (12),
16713-16725.

09 • 2O19 UnERD
AUG 13
A2: Hydrolysis of ethylene glycol vinyl ether on homogeneous and heterogeneous acid catalysts
Megan Ty1, Junnan Shangguan2, Ya-Huei (Cathy) Chin3
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto

Lignocellulose biomass is an abundant renewable feedstock with the potential to replace fossil fuels for mitigation
of greenhouse gas emissions. Brønsted acids (BA) catalyze many biomass upgrading pathways to produce valuable
bio-based products; an example is the hydrolysis of cellulose into glucose, which can be fermented into ethanol, a
fuel source. The use of homogeneous acid catalysts, in spite of offering high reactivity, has disadvantages due to
difficulty in separating the acid from reaction products. A popular alternative are heterogeneous catalysts that
possess Brønsted acid sites (BAS) on their surfaces (i.e. solid acid catalysts), as they are less corrosive and can easily
be recovered and reused. This study investigates the catalytic reactivity differences of homogeneously (oxalic acid,
acetic acid) and heterogeneously (H-ZSM-5, H-Y, Amberlyst-15) derived BA for the hydrolysis of ethylene glycol
vinyl ether (EGVE). Since the initial proton addition onto the C=C bond of EGVE is rate-limiting [1], the hydrolysis
rate of EGVE reflects the pH and acid strength of a catalyst. Reactions are carried out in an isothermal batch stirred
tank reactor system and monitored by an off-line gas chromatograph equipped with a flame ionization detector.
Initial rate measurements with the aforementioned catalysts reveal a direct correlation between pH and hydrolysis
rate. The reaction order with respect to EGVE varies between 0 to 1 across different catalysts. The comparison of
heterogeneous to homogeneous acid catalysts will aide in the development of industrial catalytic processes for
synthesizing bio-based products.

[1] A. J. Kresge and Y. Chiang. “The Hydrolysis of Ethyl Vinyl Ether. Part 1. Reaction Mechanism,” J. Chem.
Soc. B, 0, 53-57, Jan. 1 1967. doi: 10.1039/J29670000053.

UnERD
2O19 • 10
AUG 13
 A3: Efficient Extraction of Scandium From Bauxite Residue By Carbothermic Reduction Smelting
and Acid-Baking Water Leaching
Brohath Amrithraj1, John Anawati1,Gisele Azimi 1,2

Laboratory of Strategic Materials, Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Department of
1

Materials Science and Engineering, University of Toronto

Scandium is a relatively abundant element in the earth’s crust; however, its global production relies on secondary
sources, such as by-products of mining processes, due to the scarcity of concentrated scandium ore sites. The
increasing global demand for scandium [1], primarily in aluminium alloy production, requires the development of
efficient extraction methods from abundant sources such as Bauxite Residue. Bauxite Residue is a by-product in the
extraction of aluminium from Bauxite ore in the Bayer process and is rich in iron, scandium, and other rare earth
elements. In the residue, scandium is associated with iron phases. This hinders its high extraction without the co-ex-
traction of iron through conventional processes. In this study, a three-step pyro-hydro-metallurgical process was
developed involving (1) carbothermic smelting and separation of the reduced metallic iron phase and the slag phase
enriched with scandium. The slag obtained is subjected to (2) 200-400°C acid baking with concentrated sulfuric acid
followed by (3) leaching of the baked slag with water. The focus of this project is to optimize this three-step process
to maximize the extraction efficiency of scandium from bauxite residue with reduced iron impurity. The conditions
for carbothermic smelting were optimized, and the conditions for acid-baking water-leaching are being investigated
based on three factors: baking temperature, acid-to-slag ratio, and water-to-baked-slag ratio, using quadratic
response surface methodology. The proposed process shows potential to achieve high scandium extraction efficien-
cy whilst separating available iron, and thus, offers an economical and sustainable method to valorise valuable
elements available in abundant bauxite residue.

[1] " Scandium Market | Growth, Trends, and Forecast (2019 - 2024) " Mordorintelligence.com, 2019. [Online].
Available: https://www.mordorintelligence.com/industry-reports/scandium-market. [Accessed: 27- Jul- 2019].

11 • UnERD
2O19
AUG 13
A4: Interaction between organic peroxide aerosol and sulfur dioxide gas forms organosulfates in
secondary organic aerosol
Jinmyung Jang1, Shunyao Wang2, Professor Arthur Chan3
1
Department of Chemical Engineering & Applied Chemistry, University of Toronto

Particulate matter (PM) has impacts on both human health and climate change. Organic aerosol constituents large
mass fraction of fine particulate matter (PM2.5). Study revealed that up to 30% organic aerosol can be made up by
organosulfates (OS), which can be treated as the tracer for secondary organic aerosol (SOA) formation [1]. OS have
been found in many SOA from different hydrocarbon precursors, including those from biogenic and anthropogenic
emission sources. However, the formation mechanisms of organosulfates are still poorly understood due to their
complex nature. Our previous study showed that, under a controlled environment in their pure form, organic perox-
ide and sulfur dioxide (SO2) react at a significantly high rate to produce OS. The finding suggests that this reaction
can account for the missing source of OS in nature. To expand this hypothesis, this study examines more closely the
interaction between SO2 and organic peroxides in real SOA by doing chamber experiments. Peroxide content in
SOA was quantified using the iodometry assay. The uptake kinetics of SO2 onto particles with different organic
peroxides were found to be positively dependent on relative humidity. Furthermore, different types of OS formed
during the interaction between SO2 and SOA were detected by ion mobility mass spectrometry-time of flight mass
spectrometry (IMS-TOF MS).

[1] J. Schindelka, Y. Iinuma, D. Hoffmann and H. Herrmann, "Sulfate radical-initiated formation of isoprene-de-
rived organosulfates in atmospheric aerosols", Faraday Discussions, 2013.

UnERD
2O19 • 12
AUG 13
A5: Investigating pH-responsive colloidal drug aggregates for modulation of endosomal escape
Denis Qeska1, Eric N. Donders1,2, Ahil N. Ganesh1,2, Hayarpi Torosyan3, Parnian Lak3, Christian Dank4, Mark Laut-
ens4, Brian K. Shoichet3, Molly S. Shoichet1,2,4
Department of Chemical Engineering & Applied Chemistry, University of Toronto; 2Institute of Biomaterials and Biomedical Engineering,
1

University of Toronto; 3Department of Pharmaceutical Chemistry, University of California, San Francisco; 4Department of Chemistry,
University of Toronto

When some hydrophobic molecules, including chemotherapeutics, exceed a drug-specific critical concentration,
they spontaneously self-assemble into dense spherical drug aggregates. Although these colloidal drug aggregates are
nuisance artifacts in drug screening, each drug-rich aggregate has the potential to deliver high payloads of drug to
specific cells, thereby improving treatment efficacy. When these colloids are coated by a layer of antibodies, they
can be taken up by cells that express the target of the antibody. However, previous work showed that antibody-stabi-
lized colloidal drug aggregates were trapped in endosomes and lysosomes after endocytosis, thereby limiting effica-
cy. To overcome this limitation, we hypothesized that potent drugs which ionize under the acidic environment of the
endo-lysosomal pathway would destabilize the endosomes and thereby promote greater cytosolic delivery. We start-
ed with the colloid-forming chemotherapeutic, fulvestrant, and modified it with acid-responsive functional groups
with different sensitivities to acid. We then characterized how these fulvestrant analogues respond to acid and how
their in vitro anticancer activity compares to the parent drug. We also worked to develop a method for assessing
endosomal escape using fluorescently labelled endosomes and colloids, which could be tracked over time using
live-cell confocal microscopy. Our results demonstrate that otherwise inert colloidal aggregates can be made to
respond to acid, which is an important behaviour for achieving endosomal escape. Future work will investigate the
endosomal escape efficiency of the fulvestrant analogues we synthesized. This work will inform further design of
pH-responsive colloids with promise to improve efficacy of these formulations for chemotherapeutic drug delivery.
This work was supported by Amgen Scholars Program, National Institutes of Health, and National Science and
Engineering Research Council.

13 • UnERD
2O19
AUG 13
 B: Biomaterials + Biomedical
B1:Collagen hydrogels with anisotropic macropores for tissue engineering
Chen Yu Li1, Bryan Quan1, Eli Sone1,2,3
1
Institute of Biomaterials and Biomedical Engineering, University of Toronto; 2Department of Materials Science & Engineering, University
of Toronto; 3Faculty of Dentistry, University of Toronto

Collagen hydrogels are promising scaffold materials for tissue engineering, largely due to their similarity with extra-
cellular matrix [1]. Porosity is essential in tissue scaffolds; pores provide a 3D biomimetic environment with
increased surface area for cell seeding, and pore alignment directs cell migration [2]. Although methods such as
solvent casting have been used to induce porosity, it is difficult to create aligned pores with them [3]. In this study,
we present a novel method for inducing anisotropic through-thickness macroscopic pores using diffusion of ammo-
nia vapour from ammonium hydroxide (NH4OH) to neutralize acid-solubilized collagen. Pore alignment is
controlled by the top-down direction of the diffusion, and likely involves liquid-liquid phase separation to generate
collagen-rich and collagen-poor domains corresponding to bulk gel and pore, respectively. Optical Microscopy was
used to observe the porosity and images were analyzed using ImageJ, while parameters such as solution composition
and NH4OH concentration were investigated in order to control pore count and total pore area. We found that
increasing collagen and acetic acid concentrations decreased both pore count and total pore area, whereas decreasing
NH4OH concentration decreased pore count and increased total pore area. Taken together, these results suggest that
solution composition affects the thermodynamics of the system with respect to the favourability for phase separa-
tion. Conversely, NH4OH concentration affects the kinetics of phase separation and gelation, and hence the extent
of phase separation and pore formation. The novel method described in this study could be important for developing
a collagen-based scaffold for anisotropic tissue engineering applications.

[1] X. Song et al., "A Novel Human-Like Collagen Hydrogel Scaffold with Porous Structure and Sponge-Like
Properties", Polymers, vol. 9, no. 12, p. 638, 2017. Available: 10.3390/polym9120638.

[2] H. He, M. Liu, J. Wei, P. Chen, S. Wang and Q. Wang, "Hydrogel with Aligned and Tunable Pore Via “Hot
Ice” Template Applies as Bioscaffold", Advanced Healthcare Materials, vol. 5, no. 6, pp. 648-652, 2016. Available:
10.1002/adhm.201500707.

[3] N. Annabi et al., "Controlling the Porosity and Microarchitecture of Hydrogels for Tissue Engineering",
Tissue Engineering Part B: Reviews, vol. 16, no. 4, pp. 371-383, 2010. Available: 10.1089/ten.teb.2009.0639.

UnERD
2O19 • 14
AUG 13
B2: Optimization of long-term microfluidic co-cultures for modelling tumour vascularization
Kimberly Seaman1, Noosheen Walji1,2, Edmond W.K. Young1,2

Department of Mechanical and Industrial Engineering, University of Toronto; 2Institute of Biomaterials and Biomedical Engineering,
1

University of Toronto

Angiogenesis, the formation of new blood vessels from pre-existing vessels, is crucial to tumour growth. Develop-
ment of anti-angiogenic therapies has received considerable interest, but advancement is difficult due to lack of
physiologically relevant in-vitro tissue models. Microfluidics can be leveraged to observe the cancer microenviron-
ment during tumour vascularization, however optimal conditions needed to maintain viable models in microfluidic
systems have not been established. The aim of this study is to optimize a protocol to achieve long-term microfluidic
co-culture with 3D tumour spheroids and blood vessel lumens for studying tumour vascularization. Conditions to
optimize include cell seeding densities and sequence, and feeding frequencies. 3D tumour spheroids, which can
reproduce distinctive characteristics of solid tumours, are formed and subsequently loaded into the microfluidic
device. Lumens comprised of red-fluorescent protein (RFP)-expressing human umbilical vein endothelial cell
(HUVEC) monolayers are formed in the adjacent channel and co-cultured with spheroids. Co-cultures are imaged
every 24h for one week and analyzed via ImageJ to quantify cell viability. Preliminary results indicate loading
spheroids into microfluidic devices 24h before forming lumens results in higher endothelial cell viability and sprout-
ing. Co-cultures are being tested with spheroids of different types (A549 lung epithelial and MCF-7 breast cancer
cell lines) and ages (young and old) to identify changes in endothelial response, and a triple-culture system with
human normal dermal fibroblasts to enhance endothelial sprouting. Ultimately, this study will provide a method to
culture and observe tumour-vasculature systems to demonstrate the utility of microfluidic co-cultures as a
drug-screening platform for anti-angiogenic therapies.

B3: Tissue Repair with In-Vitro, Acellular, Mineralized Collagen

Zahra Nadine Kandola1

Approximately 1.4 million Canadians suffer from osteoporosis(OP); 1 in 4 women and 1 in 8 men over the age of
50. Vertebral compression fractures(VCF), a complication of OP, may present clinically with severe back pain, but
may also be asymptomatic and remain undetected. Percutaneous vertebroplasty(PV), a procedure where bone
cement is injected into a partially collapsed vertebral body, is used to relieve pain and provide mechanical stability
in cases of osteoporosis or tumor infiltration. However, the nature of these bone cements can result in local tissue
inflammation and partial tissue necrosis due to high content of methacrylate present in those cements.
Our study aims at producing a new in-vitro, acellular and mineralizable collagen gel that will be polymerizable
though natural fibrillogenesis in-situ, to form a porous and mechanically tunable scaffold that can be rapidly miner-
alized at the injury site. To develop this gel, acid soluble collagen type I molecules (as found in bones) are being
re-assembled to form 3-D scaffolds through an entropic reaction in the presence of bone minerals precursors, namely
Hydroxy-apatite nanoparticles. The morphology and structural properties of these scaffolds are being monitored by
a cross-combination of electron microscopy with Elemental Analysis(EDX), Infrared Spectroscopy and Atomic
Force Microscopy(AFM). AFM has shown successful engineering of collagen with structural properties comparable
to natural collagen. Continuous AFM use will test if the mechanical properties of the scaffold match those of natural
spinal disk. Our approach is aimed at re-engineering tissues that have lost their structure and function through func-
tional tissue engineering.

15 • UnERD
2O19
AUG 13
B4: The Use of Virtual Surgery to Predict Post-Operative Anatomy for Tetralogy of Fallot Patients
Alessia Di Nardo1, Leslie Louvelle2, Matthew Doyle1,3, Glen Van Arsdell4,5 ,Cristina Amon1,2
1
Department of Mechanical and Industrial Engineering, University of Toronto. 2Institute of Biomaterials and Biomedical Engineering,
University of Toronto. 3Division of Vascular Surgery, University Health Network, Peter Munk Cardiac Centre. 4Division of Cardiac
Surgery, University of California Los Angeles. 5Division of Cardiac Surgery, University of Toronto.

Tetralogy of Fallot (TOF) is a congenital heart anomaly characterized by four heart defects. One of these defects is
pulmonary artery stenosis. Various surgical repair strategies may be used to alleviate this stenosis, however, the opti-
mal repair for a specific patient is unknown pre-operatively [1,2]. Virtual surgery techniques, which involve virtual
modification of pre-operative geometries to represent the range of potential post-operative geometries, may enable
the prediction of the optimal repair strategy. The goal of this work is to develop a method to perform virtual TOF
repair on patient-specific geometries obtained from magnetic resonance imaging (MRI); the methodology should be
reproducible, compatible with collected patient data, and accurate. The current functioning workflow involves three
distinct steps. The first step mimics intraoperative cardio-pulmonary bypass conditions by de-pressurizing the
geometry. This is achieved with finite element analysis using ANSYS (ANSYS Inc., Canonsburg, PA, USA). The
second step creates a virtual incision and vessel patch. Using MATLAB (MathWorks, Natick, MA), the incision and
patch parameters can be specified, allowing for the creation of multiple repairs that are reproducible across various
patients. Finally, the vessel is re-pressurized to reflect the weaning off of bypass following repair. The final geome-
try can then be exported and the effectiveness of the repair can be analyzed with blood flow simulations. Moving
forward, a validation metric should be created to ensure the accuracy of the predictions made. By implementing this
methodology, it may be possible to determine the patient-specific repair strategy which optimizes hemodynamics
and potentially clinical outcomes.

[1] Starr, J. P., 2010, “Tetralogy of Fallot: Yesterday and Today,” World J. Surg., 34(4), pp. 658– 668.

[2] Chai, P. J., Jacobs, J. P., and Quintessenza, J. A., 2013, “Modern Surgical Management of Patients with
Tetralogy of Fallot,” Cardiol. Young, 23(06), pp. 905–909.

UnERD
2O19 • 16
AUG 13
B5: Characterization of exosome release profiles for use in cardiac tissue regeneration
Saifedine Rjaibi1, Karl Wagner1, 2, Milica Radisic1, 2
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Institute of Biomaterials and Biomedical Engineer-
ing, University of Toronto

Myocardial infarctions, or heart attacks, continue to be a leading global health problem, affecting over 15.9 million
people annually [1]. They occur when flow of oxygenated blood in the coronary arteries is suddenly blocked (“isch-
emia”). Heart tissue is unable to regenerate autonomously; hence, a method for repairing damaged cardiac tissue is
needed. Treatments like induced-pluripotent-stem-cell-derived cardiomyocytes have shown some success; however,
these cellular therapies present tumorigenic and arrhythmogenic concerns [2].

Extracellular vesicles (EVs), such as exosomes, are membrane-bound particles containing proteins, lipids, and
miRNAs, the last of which is postulated to promote a pro-reparative phenotype in damaged cardiac tissue [3]. EVs
are secreted from cells at varying rates depending on cell type and maturity. Therefore, this project aims to elucidate
the release profile of exosomes from the aforementioned parameters for their use in EV-mediated tissue repair.
EVs are isolated via precipitation from conditioned iPS and iCM cell culture media. The effects of 220nm filtering
are also analyzed. Nanoparticle tracking obtains the concentration and mean size of EVs. Western Blotting and
Transmission Electron Microscopy (TEM) ensures isolated nanoparticles are EVs.

Preliminary results indicated that filtration reduced the mean size of isolated cardiomyocyte samples while increas-
ing their concentration. TEM and Western Blotting assessed the membranous structure of EVs and presence of their
common markers (CD81/TSG101) respectively.

Future work includes identifying the specific miRNAs responsible for inducing this reparative phenotype and evalu-
ating their therapeutic potential in an in vitro model of ischemic cardiac injury. A better understanding of exosomes
sets the foundation for revolutionizing cardiac therapy.

17 • UnERD
2O19
AUG 13
B6: Measures of Dynamic Postural Stability for Reactive Balance in Individuals with Incomplete
Spinal Cord Injury
Mauceri S1, Lee JW1,2, Chan K2, Unger J2,3, Musselman KE2,3,4, Masani K1,2

Institute of Biomaterials and Biomedical Engineering, University of Toronto; 2KITE - Toronto Rehabilitation Institute, University Health
1

Network; 3Rehabilitation Science Institute, University of Toronto; 4Department of Physical Therapy, University of Toronto

Individuals with incomplete spinal cord injury (iSCI) often show compromised reactive balancing, resulting in
increased risk of falls. Reasons why reactive balance is compromised in the iSCI population are investigated. We
hypothesize that the individuals with iSCI would make a step at a different location than what is required to stabilize
the body. This hypothesis is tested by investigating the foot placements of these individuals during a reactive balance
task.

Twelve healthy young able-bodied (AB) individuals, fifteen age- and sex-matched AB individuals and nineteen indi-
viduals with iSCI participated in this study. The participants performed lean-and-release tests, where the participant
leaned with a tether supporting them that was suddenly released, simulating a forward fall and eliciting reactive
step(s). Multiple measures of dynamic postural stability were calculated to estimate the theoretically appropriate
foot placement during the reactive balance task, including the margin-of-stability based on the extrapolated
center-of-mass, and the misalignment between desired center-of-pressure (dCOP), a point that results in zero body
moment, and measured COP (dCOP-mCOP), a novel measure of dynamic postural stability.

The analyses are currently ongoing. We expect that the above mentioned measures can provide theoretically appro-
priate location of foot placement during the reactive task, and that these measures can reveal whether the individuals
with iSCI made steps at locations that would decrease their stability during the reactive balance task. This work can
contribute to assess the dynamic postural balance in iSCI population, which can be indirectly used to help these indi-
viduals to improve their reactive balance via rehabilitation.
of phase separation and pore formation. The novel method described in this study could be important for developing
a collagen-based scaffold for anisotropic tissue engineering applications.

C: Bioengineering
C1: Possibility of Using Proteins as Renewable Biosludge Dewatering Conditioners
Runlin Yuan1, Hamed Ghazisaidi2, D. Grant Allen3
1
Department of Chemical Engineering and Applied Chemistry, University
= of Toronto

Biosludge is the product of the activated sludge process used in wastewater treatment plants (WWTPs). Its moisture
content is usually greater than 98% and is inherently difficult to dewater due to its colloidal structure. Current major
treatments are incineration, landfilling, and composting. Because of the high water content, direct treatment using
the aforementioned methods can be energy intensive and harmful to the environment. Therefore, synthetic polymers
are often used to flocculate the sludge first, then mechanical dewatered before final treatment. However, synthetic
polymers are non-renewable materials and it is desirable to develop a more sustainable approach. Here we explore
the possibility of using proteins, which are natural polymers, as sludge conditioners to decrease or even replace the
use of synthetic polymers. Because the neutralization reaction with the negatively charged colloids’ surface in
sludge are believed to be one of the most important mechanisms as to flocculation, we chose to test the effect of lyso-
zyme, methylated lysozyme, and methylated hemoglobin, due to their surface charge and abundance. Proteins at
different dosages will first be mixed with biosludge, then gravity filtered, followed by mechanical dewatering using
crown press to simulate the dewatering process used in industry. Sludge samples will be collected at each phase to
study the change in weight, particle size, extracellular polymeric substances, and zeta potential to infer mechanisms.
From our preliminary experiments, proteins may not be capable of replacing synthetic polymers, but if successful,
this work may present an optimum protein dosage to minimize the use of synthetic polymers.
UnERD
2O19 • 18
AUG 13
C2: System engineering of Escherichia Coli for fuel gas production
Xiaotong Wang1, Yilan Liu1,2, Radhakrishna Mahadevan1,2,3,4

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2BioZone - Center for Applied Biosciences and
1

Bioengineering, University of Toronto; 3American Institute of Chemical Engineers; 4American Society for Microbiology

Renewable energy resources have always been an interest in recent decades due to the lack of crude oil on Earth.
Microbial production of gasoline alternatives has been developed to fill the deficiency of natural energy [1]. Howev-
er, the productivity has never reached the industrially accepted standard (i.e. 1 g/L/hour) [2]. In this study, a feasible
biosynthetic pathway for butane, an ideal fuel gas, has been studied and experimentally proved in Escherichia Coli.
E. Coli has been used as a microbial chassis due to its culturability [3]. To increase the productivity of butane, this
research concentrates on precursor (step-product) supplying and metabolic bottleneck releasing. Enzymes including
pyruvate formate-lyase 4 (tdcE), methylmalonyl-CoA mutase and decarbonylase (scpA and B) have been over-ex-
pressed in the pathway and tuned to supply more precursor. Besides, an in-vivo evolution system, EvolvR, will be
built up for aldehyde decarbonylase (ADse) which has been proved to be the bottleneck of the metabolic pathway
due to its low stability and catalytic efficiency [4]. EvolvR is a fused system of an error-prone DNA polymerase and
a CrispR-guided nickase. The system will be introduced to the E. Coli cells and will be guided by specific guided
RNA (sgRNA) to continuously create artificial mutations at the targeted degron of ADse while the cells are being
cultured [5]. The study will provide information on the effects of precursor availability and supply mutants to AD to
make it more stable and active. Results from the research will increase the productivity and improve the construction
of microbial platform for butane production.

[1] “Engineering microbial fatty acid metabolism for biofuels and biochemicals”. E.R. Marella, et.al. 2017. [On-
line]. Available: https://orbit.dtu.dk/ws/files/139128665/Engineering_microbi-
al_fatty_acid_metabolism_for_biofuels_and_biochemicals.pdf. [Accessed: July 26, 2019].

[2] “Microbial production of building block chemicals and polymers”. J.W. Lee, H. Kim, et.al. December 2011. [On-
line]. Available: https://www.sciencedirect.com/science/article/pii/S0958166911000383. [Accessed: July 26, 2019].

[3] “Escherichia coliminimum genome factory”. H. Mizoguchi, et.al. 2007. [Online]. Available: https://iubmb.on-
linelibrary.wiley.com/doi/epdf/10.1042/BA20060107. [Accessed: July 26, 2019].

[4] “Modular and selective biosynthesis of gasoline-range alkanes”. Sheppard, Micah J., et.al. January 2016. [On-
line]. Available: https://dspace.mit.edu/bitstream/handle/1721.1/108077/Prath-
er_Modular%20and%20selective.pdf?sequence=3&isAllowed=y. [Accessed: July 26, 2019].

[5] “CRISPR-guided DNA polymerases enable diversification of all nucleotides in a tunable window”, Shakked O.
Halperin, et.al. August 01, 2018. [Online]. Available: https://www.nature.com/articles/s41586-018-0384-8. [Ac-
cessed: July 26, 2019].
.

19 • UnERD
2O19
AUG 13
 D: Materials Science
D1: Metallic Counter Surface Interaction on Tribological Behavior of Ultrathin 2D Material
Yuhui Cheng1, Chandra Singh2
1
Institute of Material Science Engineering, University of Toronto

Two-dimensional materials are monolayered or multilayered atomic thickness structure that has weak layer to layer
interaction and amazing low friction. This results in increasing attention on the use of 2D material as lubricant.
While there has been a lot of study on interaction between graphene and other 2D materials, the study on interaction
with metal surfaces, which is more often used in real life, has been very limited. The research will look at Copper
and Titanium metal oxides and their nature of interaction such as the type of bonding as well as relative friction
behaviours when sliding with graphene and MoS2 sheets. Using Density Functional Theory modelling (DFTM), the
relative change in potential energy curves are obtained. Then interfacial shear strength is estimated using the rate of
change. Also using DFTM, the whole process is repeated with the additional consideration of van der well forces.
For copper oxide (CuO), graphene was seen to have slightly higher friction than MoS2. The oxygen rich surface for
CuO produced an order of higher friction than the copper rich surface and seems to have covalent bond type interac-
tion with the 2D sheets. The CuO surface also produced significantly higher friction than titanium dioxide surface.
For the future, other metal surfaces and their interactions with graphene and MoS2 should be studied. Also, other 2D
materials as potential candidate for lubricant should be explored in the research as well.

D2: Characterization of Microstructure of Water Atomized Steel Powder

Anagha Acharya1, Gloria Vytas1, Ali Asgarian2, Kinnor Chattopadhyay2
1
Materials Science and Engineering, University of Toronto; 2Process Metallurgy Research Labs, University of Toronto

Steel powder particles vary greatly in terms of size, porosity, and mass, in relation to production parameters such as
temperature, water flow rate, and metal flow rate. As such, their microstructure properties vary across batches and
manufacturing grades. This is not ideal for metal additive manufacturing (AM) as powder consistency is an import-
ant factor. A variety of techniques used in metal AM, such as selective laser melting (SLM) and electron beam melt-
ing (EBM) are dependent on the flowability and apparent density of the powder. This is further dependent on powder
surface condition [1], and therefore surface microstructure. Hence it is important to understand the powder micro-
structure to homogenize and improve its properties, so that AM techniques are able to realize their full potential. The
microstructures of the steel powders were imaged using a Scanning Electron Microscope (SEM), in order to charac-
terize their phase components. An etching agent, 2% Nital, was also applied to the steel samples. It excels at reveal-
ing the structure of martensite (which is a body-centered tetragonal steel that forms during quenching) and ferrite
grain boundaries in ultra-low carbon steels (ULCS) [2], such as the ones explored in this study. A surface-active
reagent, Zephiramine, will be used to reveal any dendritic structure. This will be used to calculate the solidification
time of the powder particles – arguably the most important factor in determining the morphology of the particle.
Further characterization of bismuth-tin particles produced using a lab-scale atomization chamber can potentially
prove the hypotheses around homogenizing these powders.

[1] J. Yan, Y. Zhou, R. Gu, X. Zhang, W.-M. Quach, and M. Yan, “A Comprehensive Study of Steel Powders (316L,
H13, P20 and 18Ni300) for Their Selective Laser Melting Additive Manufacturing,” Metals, vol. 9, no. 1, p. 86,
2019.

[2] G. F. Vander Voort, “IMAGING PHASES IN STEELS,” ADVANCED MATERIALS & PROCESSES, pp.
32–37, Feb. 2005.
UnERD
2O19 • 20
AUG 13
D3: Analysis of Prediction Model Accuracy for Median Particle Size of Steel Powder Produced via
Water Atomization
Gloria Vytas1, Anagha Acharya1, Ali Asgarian1, Kinnor Chattopadhyay1
1
Process Metallurgy and Research Labs, University of Toronto

The use of metal powders in manufacturing metal parts, using techniques such as additive manufacturing, metal
injection molding, and hot isostatic pressing, is increasing in popularity due to the reduction in materials waste and
ability to produce complex components. Due to this rise in popularity, the field of powder metallurgy is becoming
increasingly important since the shape, degree of homogeneity, and fineness of the powder has significant impacts
on the final product [1]. Impacts include the degree of sintering that occurs and the strength of the final product after
debinding [1].

The most common manufacturing methods of metal powders are water and gas atomization. Water atomization is a
lower-cost and less energy-intensive method, however it produces irregular particles. Models for predicting the mass
median particle diameter given powder production parameters exist, however, it is unknown which model is the
most accurate for large-scale production of ultra-low carbon steel [2] [3]. Industrial data from the water atomization
of ultra-low carbon steel was analyzed, plotted, and compared with the models to measure accuracy, with the water
flow rate to metal melt flow rate ratio as the changing variable.
Next steps include setting up a lab-scale atomizer to produce powder and collect data on water atomization to further
study the effects of other process and design parameters. Also, we will analyze the microstructure of the industrial
powders using a variety of materials analysis techniques, such as SEM, to measure solidification time and other
characteristics of the water atomization process.

[1] J. Dunkley, "Metal Powder Atomisation Methods for Modern Manufacturing", Johnson Matthey Technology
Review, vol. 63, no. 3, pp. 226-232, 2019. Available: 10.1595/205651319x15583434137356 [Accessed 26 July
2019].

[2] F. Persson, A. Eliasson and P. Jönsson, "Prediction of particle size for water atomised metal powders: param-
eter study", Powder Metallurgy, vol. 55, no. 1, pp. 45-53, 2012. Available: 10.1179/1743290111y.0000000016
[Accessed 26 July 2019].

[3] M. Pasupathy, J. Martín, A. Rivas, I. Iturriza and F. Castro, "Effect of the solidification time on the median
particle size of powders produced by water atomisation", Powder Metallurgy, vol. 59, no. 2, pp. 128-141, 2016.
Available: 10.1080/00325899.2015.1117693 [Accessed 26 July 2019].

21 • UnERD
2O19
AUG 13
D4: Effects of Electrolytes to Adsorption and Diffusion of Lithium Polysulfides over Black Phos-
phorene based Cathode for Lithium-Sulfur Batteries
JongHyun Sa1, Sankha Mukherjee1, Chandra Veer Singh1,2
1
Department of Materials Science and Engineering, University of Toronto; 2Department of Mechanical and Industrial Engineering, Univer-
sity of Toronto

Lithium-Sulfur (Li-S) batteries are considered to be next generation energy storage devices as they are low cost and
can operate at extreme temperatures [1]. However, the polysulfide (PS) shuttle effect, where lithium-PS (LiPS)
species are reduced at the negative electrode surface, occurs during the charge and discharge process [2]. Such effect
decreases the efficiency of batteries and stimulates the electrolyte decomposition which causes battery failure [1].
Recently, experimental studies demonstrated black phosphorene (α-P) to be a promising 2D cathode material for
effectively immobilising LiPS [3]. By applying α-P in Li-S batteries, it is possible to control the PS shuttle. Nonethe-
less, effects of an electrolytic environment to the chemical interactions between α-P and LiPS are yet to be found.
Herein, density functional theory (DFT) calculations are employed computationally to evaluate the electrostatic
interactions between α-P and LiPS with five different dielectric constants (εr), which correspond to toluene, tetrahy-
drofuran, acetone, dimethylsulfoxide, and ethylene carbonate environments. Li2S and Li2S2 are analyzed as insolu-
ble LiPSs, and Li2S4, Li2S6, and Li2S8 are analyzed as soluble LiPSs. Our results demonstrated that increasing the
dielectric constant of the electrolyte resulted in a decrease in the strength of the binding energy (EB) of LiPSs on
α-P. Next steps include estimating energy barriers associated with the migration and decomposition of larger PSs
over α-P. Our results not only provide insight into the interactions between α-P based cathodes, PSs and electrolytes,
but also serve as a starting point for rational cell design of Li-S batteries.

[1] H. Zhang, X. Li, and H. Zhang, Li-S and Li-O2 Batteries with High Specific Energy: Research and Develop-
ment. Singapore: Springer Singapore, 2017.

[2] A. N. Mistry and P. P. Mukherjee, “‘Shuttle’ in Polysulfide Shuttle: Friend or Foe?,” The Journal of Physical
Chemistry C, vol. 122, no. 42, pp. 23845–23851, Oct. 2018.

[3] J. Zhao, Y. Yang, R. S. Katiyar, and Z. Chen, “Phosphorene as a promising anchoring material for lithi-
um–sulfur batteries: a computational study,” Journal of Materials Chemistry A, vol. 4, no. 16, pp. 6124–6130, 2016.

UnERD
2O19 • 22
AUG 13
D5: Investigating Microstructural Radiation Damage in Tungsten for Nuclear Reactor Design
Apurv Kaushik1, Eric Nicholson1,2, Professor Chandra Veer Singh1,3
1
Materials Science and Engineering, University of Toronto; 2University of Toronto Institute of Aerospace Studies; 3Mechanical and Indus-
trial Engineering, University of Toronto

Due to its outstanding melting point and strength, tungsten is a candidate material for plasma-facing components in
tokamak fusion reactors. Understanding its behavior under reactor conditions is vital to nuclear fusion reactor
design. Nuclear energy accounts for 60% of Ontario’s power generation. Thus, developing cleaner, more efficient
reactors enormously benefits industry, society, and the environment. Energetic particle irradiation creates defects in
the regular arrangement of tungsten atoms. These include vacancies, or missing atoms, and interstitials, where
smaller atoms such as hydrogen occupy the spaces between tungsten atoms. The effects of these radiation-induced
defects on the microstructure of tungsten are not well understood. Under surface-sensitive nanoindentation testing,
radiation damage in tungsten is indicated by the elimination of pop-in behavior: the sudden onset of higher-dimen-
sional defects called dislocations under loading. This investigation aims to clarify the impact of radiation-induced
microstructural defects, by observing the decrease in pop-in behavior under nanoindentation. Nanoindentation of
irradiated (tungsten with postulated defect concentrations) and non-irradiated samples (pristine tungsten) is simulat-
ed using classical molecular dynamics (MD). Dislocations are analyzed in post-processing software, and the thresh-
old energy required for a pop-in to occur is calculated from the corresponding load-displacement curve. Preliminary
results indicate dislocation formation in non-irradiated samples. It is predicted that irradiated samples will demon-
strate reduced pop-in behavior, and that a correlation between the pop-in threshold and defect concentration may
emerge, suggesting an optimal defect concentration for future experimental studies. These results will provide
insight into the microstructural effects of radiation on tungsten, potentially aiding fusion reactor design.

D6: Investigating the effects of graphene size and cathodic material loading on the performance of
aluminum-ion batteries using a graphene nanoplatelet based cathode
Tony Dong1, Kok Long Ng1, Gisele Azimi1,2

Department of Materials Science and Engineering, University of Toronto; 2Department of Chemical Engineering and Applied Chemistry,
1

University of Toronto

Our growing need for cheap and efficient energy storage in support of sustainable energy systems has encouraged
extensive research into next-generation batteries. Favored for aluminum’s abundance, high specific/volumetric
capacity and ease of handling, the aluminum-ion battery (AIB) is promising to fulfill such a need. Recent research
of AIBs has focused on employing graphitic materials as the cathode, and ionic liquids containing chloroaluminate
anions (AlCl4-) as the electrolyte. While it is known that the physical dimensions of graphitic particles can affect the
intercalation kinetics of AlCl4- and hence battery performance, there is a lack of quantitative investigations relating
this size effect and associated cathodic parameters like material loading to battery performance metrics. To elucidate
this relationship, the present study utilizes graphene nanoplatelets (GNPs) of known lateral sizes and thicknesses as
the AIB cathode and approaches the investigation statistically through the response surface methodology (RSM)
with a pentagonal design. Cathodes are fabricated with specific combinations of GNP size and material loading (g
GNP/cm2) determined by RSM and tested in pouch cells galvanostatically to evaluate critical performance parame-
ters including specific capacity, coulombic/energy efficiency and average discharge voltage. Preliminary results
show that specific capacity can be improved by using GNP with larger lateral size and thickness. Further correlations
are expected to be elucidated as data analysis progresses. These results will not only offer additional insight into the
intercalation mechanism of AlCl4- at AIB cathodes but also serve as a guide for the design of high performance
AIBs that can facilitate storage of sustainable energy.

23 • UnERD
2O19
AUG 13
D7: High throughput analysis of water splitting using two-dimensional transition metal dichalco-
genides
Daniel Persaud1, Shwetank Yadav2, Chandra Veer Singh3,
1
Department of Material Science and Engineering, University of Toronto; 2CME Lab

As climate change continues to worsen, finding an alternative to burning fossil fuels is incredibly important. While
the combustion of hydrogen is more efficient, the hydrogen generation process is very difficult thus hindering its
viability as an alternative. The purpose of this work is to investigate the potential of hydrogen generation via water
splitting using transition metal dichalcogenides (TMDs). Since experimentally evaluating hundreds of TMDs would
be very costly and time consuming, computational simulations are utilized. This is done by first modeling each TMD
and then relaxing the model in Quantum Espresso so the rest of the simulation is carried out as it would in reality.
Then a water molecule is placed above the surface of each TMD to be relaxed again and using Density Functional
Theorem, the energy required to adsorb water is calculated. Currently, the water simulations are being carried out,
but as of now, -H has reported the lowest adsorbance energy. Once the absorbance energy of water is computed for
each TMD, they can be compared in order to rank each TMDs ability to adsorb and thus split water. In the future,
machine learning will be utilized to find a correlation between a property of the TMDs and its ability to split water.

D8: Surface Engineering of Carbon Electrodes with Pseudocapacitive Layers for Electrochemical
Capacitors
Shaheer Siddiqui1, Jeanne N’Diaye1, Keryn Lian1
1
Department of Materials Science and Engineering, University of Toronto

Electrochemical capacitors (EC’s) offer a cleaner, more efficient future in the energy storage field. They have faster
charge/discharge capabilities and longer cycle lives. The electrode components for EC’s rely on carbonaceous mate-
rials such as carbon nanotubes (CNT) which exhibit ideal capacitive behaviour, represented by equal charge and
discharge. But the charge accumulation on the electrode is too minimal for practical applications. Applying thin
layers of pseudocapacitive materials such as polyoxometalates (POM’s) via a layer-by-layer (LbL) deposition to
induce surface redox reactions can effectively improve the charge accumulation of CNT by up to 10 times. However,
the coatings may shift the capacitive profile away from ideal conditions. Furthermore, the intermediate polymer
layer used to anchor the POM’s may add resistance to the electrode, limiting the suitability of the EC. In this study,
the surface chemistry of CNT was modified by different combined electroactive conductive polymers (CP’s) and
POM’s to obtain an ideal capacitive profile. Recently modified electrodes, such as CNT coated with polyluminol
(CpLum) and phosphomolybdic acid (PMo), were analyzed through cyclic voltammetry (CV) and showed promis-
ing results through a better capacitive profile and higher current. The resistance of these electrodes will be tested
using electrochemical impedance spectroscopy (EIS). The resulting CV curves and EIS spectra will provide guid-
ance towards optimizing future combinations of CP’s and POM’s so that an ideal capacitive profile can be achieved.
The improved capabilities of EC’s will make it a solution for future energy needs, in sectors such as wearable tech-
nology and public transportation.

UnERD
2O19 • 24
AUG 13
D9: A Literature Review on the Electrochemical Isotope Separation of Key Nuclear Materials
Jackson Ngo1, Robert Alicandri1, and Kinnor Chattopadhyay1
1
Process Metallurgy Research Labs, University of Toronto

Electrochemical separation of elemental isotopes has recently been theoretically described and demonstrated for a
variety of transition metals such as Fe [1] and Zn [2]. This work outlines the industrial need of isotopically upgrad-
ing several elements that are otherwise uneconomical or require highly specialized equipment to separate by current
methods. These include, but are not limited to, centrifugation, laser/electromagnetic separation, chemical or organic
solvent extraction, and thermal diffusion, among others. [3] Isotope separation is a typical process for nuclear-relat-
ed applications where the varying characteristics of a selected isotope can severely alter a material’s performance in
given applications. An example of this being differing values of neutron cross-sections affecting the susceptibility
of a material absorbing incoming neutrons. [4] This work will focus on elements relating to the energy sector, partic-
ularly 90Zr, 6Li, 7Li, and 235U, where there has not been substantial research effort on the development of new
separation techniques in some time. The state-of-the-art isotope separation methods for each element will be
discussed with a comparison of each method’s benefits and drawbacks. It will conclude with a suggested research
plan in addition to potential chemistries to assess the technical and commercial viability of electrochemical isotope
separation.

[1] J. Black, E. Young and A. Kavner, "Electrochemically controlled iron isotope fractionation", Geochimica et
Cosmochimica Acta, vol. 74, no. 3, pp. 809-817, 2010.

[2] J. Black, S. John and A. Kavner, "Coupled effects of temperature and mass transport on the isotope fraction-
ation of zinc during electroplating", Geochimica et Cosmochimica Acta, vol. 124, pp. 272-282, 2014.

[3] M. Ragheb, "Isotopic Separation and Enrichment", University of Illinois Urbana, 2019. [Online].Avail-
b e t t p : / / w w w . r a g h e b . c o / N -
PRE%20402%20ME%20405%20Nuclear%20Power%20Engineering/Isotopic%20Separation%20and%20Enrich
ment.pdf.

[4] DOE Fundamentals Handbook, Nuclear Physics and Reactor Theory, vol 1, 1st ed. Washington, D.C.: U.S.
Dept. of Energy, 1993, p. 111. Available: https://www.standards.doe.gov/standards-docu-
ments/1000/1019-bhdbk-1993-v1

25 • UnERD
2O19
AUG 13
D10: Investigating the Nanoscale Damage of Silicon Carbide under Hydrogen Bombardment
Julien Zhu1, Eric Nicholson2, Chandra Veer Singh3
1
Department of Mechanical and Industrial Engineering, University of Toronto; 2Department of Materials Science and Engineering, Univer-
sity of Toronto; 3Department of Materials Science and Engineering, University of Toronto

Tokamaks are the most developed nuclear fusion devices. Hydrogen isotopes, necessary for fusion, bombard the
container wall with energies up to several hundred electron-volts (eV) [1]. To ensure safety, understanding the
evolution of the properties of the material of the reaction chamber under irradiation is paramount. A candidate mate-
rial is silicon carbide (SiC) which is advantageous for its low tritium retention and high thermal tolerance [2]. How-
ever, experiments have suggested that radiation-induced displacement damage may be caused by lower energies
than what is expected under head-on elastic collision theory - a concern in maintaining the integrity of the container
wall [3].

To explain the underestimated damage, we aim to understand the effects of carbon surface impurities mediating the
energy transfer from incident particles. The mediation may impart greater energy to the material than direct bom-
bardment alone. We will simulate hydrogen bombardment above the free surface of a 10x10x10 unit cell box equili-
brated at 150K via LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) on Niagara, Canada’s
largest supercomputer.

The energies of the incident hydrogen atoms will be compared with the secondary knock-on atom, either silicon or
carbon, to create distributions of the energy imparted. Currently, the Tersoff/Ziegler-Biersack-Littmark interatomic
potential produces threshold displacement energies between 16 eV and 55 eV for the carbon sublattice and between
22 eV and 60 eV for the silicon sublattice. These values are the baseline for understanding the radiation damage of
SiC and the energy distributions to be developed will expand on the threshold displacement energy models.

[1] Y. Ueda, J. Coenen, G. D. Temmerman, R. Doerner, J. Linke, V. Philipps, and E.

Tsitrone, “Research status and issues of tungsten plasma facing materials for ITER and beyond,” Fusion Engineer-
ing and Design, vol. 89, no. 7-8, pp. 901–906, Oct. 2014.

[2] R. A. Causey and W. R. Wampler, “The use of silicon carbide as a tritium permeation
barrier,” Journal of Nuclear Materials, vol. 220-222, pp. 823–826, Apr. 1995.

[3] A. Krasheninnikov, E. Salonen, K. Nordlund, J. Keinonen, and C. Wu, “Tight-

Binding Atomistic Simulations of Hydrocarbon Sputtering by Hyperthermal Ions in Tokamak Divertors,” Contribu-
tions to Plasma Physics, vol. 42, no. 2-4, pp. 451–457, 2002.

UnERD
2O19 • 26
AUG 13
D11: Structural Analysis of Corneal Nano-nipple Arrays in Moth Eyes
Tianyi Lyu1, Uwe Erb1
1
Department of Materials Science and Engineering, University of Toronto

The corneal protuberance (called “nano-nipple”) arrays in the compound eyes of moths/butterflies have been studied
as a template for anti-reflective structures. Past research argued that the ordered arrays were hexagonally
close-packed, and thus the arrays were treated as continuous p6mm plane groups [1][2]. Recent reviews showed that
there were various degrees of deviations from the p6mm structure, depending on the area of the eye and species. Past
research also measured several important parameters such as nano-nipple size and lattice parameter length, yet there
was a little comparison among species [1][2]. This research aims to fill the gaps by introducing a new parameter:
nano-nipple size to lattice parameter ratio. The hypotheses are: (1) the ratio should be a species-specific constant;
(2) a higher ratio would correspond to arrays with high p6mm resemblance while a lower one would correspond to
arrays with high p2 resemblance. To validate, nano-nipple arrays of three moth species were characterized using
scanning electron microscope (SEM) and crystallographic principles were applied to perform the analysis. The
preliminary results showed that, firstly, the ratio was a constant within each specimen, but could vary within a
species, suggesting there were more determiners such as gender, age, etc. Secondly, the relationship between the
ratio and the plane group resemblance meet the hypothesis, but there were slight regional differences within each
specimen. The next step of the work was to investigate both the inter-species and intra-species differences by using
simulation software based on the reaction-diffusion mechanism proposed by Alan Turing.

[1] K. Lee, Q. Yu and U. Erb, "Mesostructure of Ordered Corneal Nano-nipple Arrays: The Role of 5–7 Coordi-
nation Defects", Scientific Reports, vol. 6, no. 1, 2016. Available: 10.1038/srep28342.

[2] K. Lee and U. Erb, "Remarkable crystal and defect structures in butterfly eye nano-nipple arrays", Arthropod
Structure & Development, vol. 44, no. 6, pp. 587-594, 2015. Available: 10.1016/j.asd.2015.08.007.

27 • UnERD
2O19
AUG 13
 E: Electrical
E1: Optimizing High-Level Synthesized Circuits with Register Insertion and Profiling Method

Kexin Li1 ,Yu Ting Chen2,3, Jin Hee Kim2,3 and Jason Anderson2,3
1
The Edward S. Rogers Sr. Department of Electrical & Computer Engineering, University of Toronto

Computer hardware design is more difficult and time-consuming than software because it is more low-level than
pure software. High-level synthesis (HLS) tools are highly demanded by field-programmable gate array (FPGA)
companies since they eliminate the complexity of hardware designs. LegUp is an HLS tool that compiles C/C++ into
hardware description languages (HDL), such as Verilog. Although LegUp can translate software languages into
HDL flexibly, it did not support Stratix10 which is the latest Intel FPGA incorporating bypassable registers.
The purpose of this research is to incorporate Stratix10 support into LegUp to optimize the speed for register-trans-
fer level (RTL) designs without drastically increasing clock latency. The primary metric is the maximum clock
frequency (Fmax), indicating the speed of the design. Register insertion is implemented to increase Fmax by break-
ing down lengthy datapaths. However, the trade-off of Fmax improvement is the increase of clock latency, which is
the second metric. Another metric is wall-clock time (latency/Fmax), and in order to achieve lower wall-clock time,
Fmax should be improved with a modest increase in latency.

Sanity checks are conducted on 12 benchmarks. Simulation results conclude the baseline with 224.78MHz Fmax,
15,755 latency and 70.13μs wall-clock time. Final results show that Fmax increase to 310MHz with 33,371 latency
and 107.62μs wall-clock time when inserting one register after all instructions, and inserting three registers increase
Fmax to 387.7MHz with 60,726 latency and 156.63μs wall-clock time. Although Fmax has significant improve-
ment, latency impacts wall-clock time. To minimize wall-clock time, registers are inserted smartly after instruction
blocks. The instructions are selected by the profiling algorithm which considers the number of executions and
cycles. The implementation of the profiling algorithm not only improves Fmax from 224.78MHz to 278.13MHz but
also decreases wall-clock time to 64.88μs from 70.13μs. Those results indicate that LegUp is capable of producing
faster circuit designs without sacrificing wall-clock time. The results can potentially be incorporated into future HLS
commercial tool development to provide faster synthesized circuit designs for digital circuit designers and compa-

UnERD
2O19 • 28
AUG 13
E2: Camera motion removal in egocentric videos for more robust hand use evaluation in individu-
als with SCI
John Salama1, Andrea Bandini2,3, José Zariffa1,2

Institute of Biomaterials and Biomedical Engineering, University of Toronto; 2Toronto Rehabilitation Institute (TRI), University Health
1

Network (UHN)

Individuals with cervical spinal cord injuries (SCI) experience upper limb weakness or paralysis. The development
of effective new rehabilitation interventions for improving upper limb function is limited by the inability to directly
evaluate hand use during daily activities at home. Wearable cameras and computer vision allow extracting useful
information about the hand movements and interactions of individuals with SCI, for upper limb function assessment
in a daily setting. Because egocentric cameras are mounted on the patient’s head, background motion due to head
movements affects the robustness of algorithms for hand-object interaction detection. The aim of this project is to
develop an adaptive method that suppresses background motion in egocentric videos of hand use.
First, the hand regions from two consecutive frames are masked. Subsequently, features from the background region
are extracted and matched between the two frames to estimate a newly warped image with suppressed background
motion while preserving the hand appearance and movement. Optical flow represents the magnitude and direction
of pixel movements between two consecutive frames. Comparing the optical flow histogram mean with and without
the algorithm is an objective measure of the design’s performance.
In a preliminary evaluation on 1050 frames from a healthy participant, the algorithm successfully reduced the back-
ground optical flow histogram mean by 82%, while maintaining the hand optical flow. Further research about frame
features to be detected has the potential to significantly increase the accuracy of the estimated warped image thereby
improving the robustness of the algorithm.

E3: Neural Interface System for Real Time Microelectrode Array Adaptive Neurostimulation
Adam Gierlach1,2, Gerard O’Leary1,2, Roman Genov2, Taufik A. Valiante1,3,4
1
Krembil Research Institute, University Health Network; 2Dept. of Electrical and Computer Engineering, University of Toronto; 3Institute
of Biomaterials and Biomedical Engineering, University of Toronto; 4Division of Neurosurgery, Dept. of Surgery, University of Toronto

The ultimate tool in neuroscience would offer the ability to record from every cell and adaptively stimulate each
neuron based on inferred states. Microelectrode arrays (MEAs) enable high resolution in-vitro interaction with
animal, human, and artificially cultured neural tissue, allowing for the detailed analysis of neural mechanisms that
underly cognition, memory, and neuronal computation. However, there are two critical bottlenecks of existing MEA
systems: 1) the latency between recording, processing the data, and stimulation; 2) inflexible, predefined stimula-
tion. Presented here is a microelectrode neural interface system (µNIT) which integrates hardware accelerators for
real-time analysis of neuron-level activity, and adaptive generation of responsive electrical stimuli. This enables
custom stimulation within 75 milliseconds, a key threshold for neuroplastic change to occur. Inferred states are used
to adaptively control programmable waveform generators which create per-electrode stimuli at a 22.3 kHz sample
rate. The processing system is demonstrated in-vitro with clinically resected human brain tissue. Signals are digi-
tized at the source using modular custom printed circuit boards that can support any MEA and scalable to any
number of channels. The hardware accelerators are also modular for varying processing requirements. The µNIT
system costs ~$3000, ~20x less than existing systems. µNIT takes steps towards the ability to stimulate and record
from every neuron in the brain with high-performance, low-latency processing in the loop which could revolutionize
neuroscientific research. Such systems could be used for neurostimulation parameter space exploration to find more
effective closed-loop treatment approaches for disorders such as epilepsy and Alzheimer’s disease.

29 • UnERD
2O19
AUG 13
E4: Metaveillance and Metaveillography: Auditing Surveillance Cameras Using Autonomous
Drones
Yi Xin Xiang1, Steve Mann1, Cayden Pierce1, Jesse Hernandez1, Qiushi Li1, Beicong Zheng1
1
MannLab Canada, 330 Dundas Street West, Toronto, Ontario, M5T 1G5

Surveillance cameras are used to check for illegal activities and protect properties. It is difficult to look behind the
dark shields covering cameras and to visualize what direction the cameras are looking in. We propose metaveillance,
i.e. the sensing-of-sensing or auditing of surveillance, where different colours of light represent regions in the field
of view of a camera [1]. Auditing surveillance cameras can be used to identify leaks in the surveillance coverage.
The objective is to design a set of procedures that outputs an intuitive metaveillance graph for people to understand
the surveillance efficacy. The design should accurately depict the camera’s field of view and be portable and applica-
ble to cameras implemented at any location.

The current design involves using an autonomous drone, which flies around the camera in a cone shape. A program
processes the camera frame and determines whether the drone is in or out of the field of view. Based on the output,
a LED light on the drone will change colour to indicate an extent of veillance that ranges from red through green.
We have currently obtained long exposure metaveillance photos and recorded drone position and camera output
data. Next steps involve improving drone paths for better visualization and processing data to check the accuracy of
the field of view measurements against camera specifications. A scientific rating may be developed regarding the
metaveillance flux density coverage, and insurance companies can use this to evaluate the level of security and thus
determine an insurance premium.

[1] S. Mann. “Phenomenal augmented reality: Advancing technology for the future of humanity,” IEEE Consumer
Electronics, pp. cover + 92-97, October 2015.

UnERD
2O19 • 30
AUG 13
F: Mechanical & Industrial
F1: Predicting droplet penetration time into a porous bed using an open-source pore network
model.
Joanna Kolodko1, Jordan Bouchard2, Sanjeev Chandra3
1
Department of Mechanical and Industrial Engineering, University of Toronto

Experimentally studying how a droplet penetrates porous materials is difficult. The impact on the surface can be
observed, but penetration below the surface is obscured by the opaque media. We can counter this issue by perform-
ing computational fluid dynamics simulations, but such simulations are computationally expensive and often require
specialized training. Pore network models simplify a porous media into voids and connective tubes. They can
predict the flow of liquid through a porous media but have not yet been used to model single droplet penetration. We
propose expanding an existing open-source pore network model, PoreSpy, using the programming language Python,
to create a readily available software that predicts droplet penetration time and extent of saturation of single drop
impacts on porous media while reducing computational requirements. A physical model to describe the contribution
of the droplet's inertia to the penetration rate will be added to the software's existing capillary driven penetration
model. The porous media used in the software will be constructed from microCT scans of porous material from
previous single droplet experiments to validate the model directly against the experimental data. The expanded
software will allow pore size, porosity, and droplet impact velocity to be varied to a larger degree than can be done
experimentally; thus, increasing our understanding of the physics that control drop penetration into porous material.
This work will provide the research community with an affordable open-source method for predicting penetration
time and extent of saturation of single drop impacts on porous media.

F2: A Markov Decision Process and Model-Free Reinforcement Learning for Inventory Manage-
ment
Jiahua Chen1, Aloagbaye Momodu2, Chi-Guhn Lee3
1
Department of Mechanical and Industrial Engineering, University of Toronto

Big food manufacturing companies (e.g. Nestle) have to fulfill orders from many clients, but sometimes these orders
exceed the company’s inventory capacity. Usually, the company will choose to allocate more inventory to important
clients (e.g. Walmart) while not meeting orders from other clients (e.g. Sysco). As a consequence of under-fulfill-
ment, some pre-specified penalty value will be applied to the company. Currently, there are no suitable RL applica-
tions to the inventory management problem. The aim of this study is to reduce penalty values paid by modelling the
problem as a Markov Decision Process (MDP) and optimizing using a Model-Free Reinforcement Learning (RL).
Elements in MDP are state, action, reward, and discount factor. State considers the order quantity of product “p” for
retailer client “r” at timestamp “t” and the current inventory level of “p”; actions are the fill rates for each required
order quantity; rewards are the penalty amounts pre-specified in the vendor agreements with each client; and a
discount factor of 0.95 was selected. All the data is provided by our manufacturing company. To confirm that alloca-
tion strategies suggested by the algorithm result in the least penalty, we will compare them with historical data to see
if they decrease the penalty cost. The final deliverable will be an application with inputs for the elements of the MDP
and outputs of an optimal inventory allocation for each client. This study will provide insight on inventory manage-
ment for big food manufacturing companies, reduce money charge on order under-fulfillment, and thereby increase
earnings.

31 • UnERD
2O19
AUG 13
F3: Improving acoustic properties of polypropylene foams through graphene content and expan-
sion ratio
Eva (JiaYi) Liu1, Yun-seok Jun1, Chul B. Park1, Jianfei Pan1
1
Microcellular Plastics Manufacturing Laboratory, Department of Mechanical and Industrial Engineering, University of Toronto

Sound pollution has become a growing problem with the rise of noise in everyday living. Long term exposure to the
level of noise typically found in cities can lead to an increased risk of cardiovascular illnesses, hearing loss and
diabetes.[1] Currently, materials commonly used for sound absorption are polyurethane based, which possesses
difficulties during recycling and produces a poisonous gas when incinerated. Thus, thermoplastics such as polypro-
pylene (PP) are in demand as a more environmentally friendly and safe alternative. [2] This project focuses on the
introduction of graphene into foamed PP to increase sound absorption properties, which is due to graphene’s
well-connected hierarchical cavities. It will also increase the mechanical properties and electrical and thermal
conductivities of the material. [3] The samples tested are injection molded PP foams that have expansion ratios of
three to nine and graphene contents of one to six percent. They are perforated prior to testing so that sound may
penetrate the dense outer layer that lacks the presence of sound absorbing open-cells. The materials’ absorption coef-
ficient and transmission loss are measured across a frequency range of 1000 to 6300 Hertz using an impedance tube.
Preliminary results show a marked improvement in absorption coefficients between neat samples and samples with
higher graphene content of the same expansion ratio. Further plans include utilizing scanning electron microscope
imaging and pycnometer open-cell content measurements to understand the effect of graphene on cell morphology,
as well as the inconsistencies in trend for the lower graphene content samples.

[1] University of Adelaide, "Superior noise control using graphene", Phys.org, 2019. [Online]. Available:
https://phys.org/news/2019-02-superior-noise-graphene.html.

[2] J. Davoud, "Injection Molded Thermoplastic Foams for Sound Insulation", Ph.D, University of Toronto,
2015.

[3] B. Lu et al., "High performance broadband acoustic absorption and sound sensing of a bubbled graphene
monolith", Journal of Materials Chemistry A, vol. 7, no. 18, pp. 11423-11429, 2019. Available: 10.1039/c9-
ta02306b.

UnERD
2O19 • 32
AUG 13
F4: Pricing American Put Options through Reinforcement Learning Algorithms
Kai Yang Chen1,2, Yuri Lawryshyn1
1
Centre for Management of Technology and Entrepreneurship (CMTE), University of Toronto; 2Division of Engineering Science, University
of Toronto

Options are among the most actively traded derivatives in financial markets. The two most common option types are
European and American. Pricing of European options can be solved analytically using the Black-Scholes equation.
But for American options, because they give the owner the right to exercise at any time (i.e. “continuous time”
instrument), analytical solutions are not available, and instead, require numerical methods.

With the success of machine learning algorithms, especially reinforcement learning, in recent years, there is an
emerging interest in implementing them in financial markets. Although there is an abundant number of published
research papers predicting stock market movements using deep neural networks, few are focused on using reinforce-
ment learning to price derivatives. Thus, applying reinforcement learning algorithms for American option pricing is
a nuance area for exploring.

Stock price data is generated via Monte Carlo simulation with discrete time steps and assumed volatility and
risk-free return rate. To simplify the learning problem, user can define time(s) when the option can be exercised.
Then, the reinforcement learning agent travels through all paths in the simulation and learns an optimal exercise
strategy. Using the learned strategy, one can price an option with multiple exercising points. Such price is then com-
pared to an analytical solution via binomial tree method under risk-neutral probabilities. Pushing the exercising
point into a continuous scheme, one can price an American option. Expanding this methodology, one can add more
variables and extend the problem space into a real options framework.

F5: Nestle’s Deduction Claims Classification Using Clustering and Classification

Jiaru Li1, Ruiyang Li1, Aloagbaye Momodu1, Chi-Guhn Lee1
1
Dynamic Optimization & Operations Management Lab, Department of Mechanical and Industrial Engineering, University of Toronto

As the world's largest food company, Nestle faces the problem of receiving deduction claims worth about 25 million
dollars every year. When underpayment happens, Nestle decides whether to investigate and the order of investiga-
tion. Despite the large number of fines, the time and money costly manual process is used currently, instead of an
automatic process to determine the success rate. The aim of the design is to create an automated process to increase
efficiency, and to bring economic benefits. Clustering, one of the machine learning techniques, will be used to classi-
fy each claim into 3 groups (Green, Yellow or Red) more accurately and optimize the decision-making process. The
green indicates the claim is hard to turn down or the disputed amount is small, the red represents the easiness of turn
down or the significance of the amount, while the yellow is the intermediate level. Clustering is quick and efficient
which makes it useful for large data sets. Classification method will segment data into two categories, either success
or failure. The proposed models can help Nestle identify which group any claim belongs to. The models can be
tested by the comparison between the result and the real-world scenario, the decrease in the total fines paid as com-
pared to the amount paid without the models, as well as the time taken to get the classification output. For future
implementation, a visualization in PowerBI will be created to make it accessible to all employees.
.

33 • UnERD
2O19
AUG 13
F6: Design of Ambient Augmenting Technologies for Individuals Living with Cognitive Decline
and Dementia
Michelle Li1, Farzad Nejatimoharrami2,3, Mark Chignell1,2,3
1
Interactive Media Lab, University of Toronto

Cognitive decline and dementia encompasses a variety of incurable and slow progressing brain diseases. As the
population ages and the average lifespan increases, it is expected that caring for individuals with cognitive decline
will become an increasingly significant societal issue. For individuals with dementia living in long term care (LTC)
facilities, the current style of care often results in under-simulation and social isolation [1]. Current treatment is often
limited to sedative antipsychotic drugs or behavioural management techniques resulting in increased caregiver
burden [2]. A gap exists for ambient augmenting technology (AAT) for LTC facilities to stimulate patients with
cognitive decline while providing data for caregivers to track patient cognitive status. To provide a solution, a collec-
tion of technology based activities will be designed incorporating mental and physical activity with a tailored user
experience for older aged persons. Activities being designed include a bicycle travel simulator, as well as variations
of the classic whack-a-mole game intended to measure response inhibition, spatial attention, and cognitive speed
and flexibility. Web and mobile versions of the applications will be implemented, with patient usage data made
accessible to caregivers through a dashboard application. Evaluating the AAT’s ability to track cognitive status data
will be based on the technology’s ability to record and communicate a wide scope of measurements to caregivers.
Data gained from user experience of LTC patients such as time spent using the AAT and results from a user survey
will be analyzed to evaluate the activities’ ability to provide engaging stimulation for patients.

[1] A. Wilkinson, J. Blumenthal, M. Kanik, The Iterative Design of Ambient Augmenting Activities for People
Living with Dementia”. University of Toronto, 2018.

[2] M. Chignell, A. Wilkinson, T. He, F. Nejatimoharrami, O. Shevchenko, Behavioural Management in Demen-

tia Using Innovative Devices and Systems. University of Toronto, IT University of Copenhagen, University of
Edmonton, 2018.

UnERD
2O19 • 34
AUG 13
F7: Optimal Liquidation Strategies Under a Transient Price Impact Model
Matthew Reiter1, Masaaki Fukasawa2
1
Department of Engineering Science, University of Toronto; 2Graduate School of Engineering Science, Osaka University

For an agent looking to liquidate an inventory of shares, the optimal trading policy returns the greatest liquidation
wealth while striking a balance between price impact risk and price uncertainty risk. Price impact risk culminates in
transactions costs when trading is executed too quickly, stripping the market of liquidity and thus stipulating a penal-
ty under the laws of supply/demand. In contrast, price uncertainty risk is detrimental for strategies which trade too
slowly, since the agent forfeits accuracy in predicting the expected value of the share-price in the future. The aim of
this research is to characterize the optimal inventory liquidation policy that a risk-adverse agent must follow to max-
imize her expected terminal wealth. The underlying model considers a block-shaped limit order book with an
accompanying bid/ask price process that affords directional sensitivity to trading volumes and demonstrates rever-
sion to a fundamental, unaffected price process. In characterizing optimality, the solution method strays from the
typical approach taken in control theory – solving the Hamilton-Jacobi-Bellman equation – where instead it is
shown that the convex value-functional may be discretized and solved as a Quadratic Program. A supporting analy-
sis using techniques from Variational Calculus is presented to justify the discretization approach. The results from
this work admit insight into economic factors such as market depth, resilience, tightness and bias. Continued work
under this framework may be motivated by expanding the dimensionality of the model to cover multiple assets or
by introducing a game-theoric approach to study a Nash equilibrium for a market with competing agents.

35 • UnERD
2O19
AUG 13
 G: Aerospace
G1: A comparative evaluation of energy- and entropy-stable Discontinuous-Galerkin and Summa-
tion-by-Parts methods
Yewon Lee1, Masayuki Yano2, David W. Zingg2
1
Division of Engineering Science, University of Toronto; 2University of Toronto Institute for Aerospace Studies

High-order methods can solve problems in computational fluid dynamics with high accuracy at low cost [1]. How-
ever, the adoption of high-order methods such as Discontinuous-Galerkin (DG) and Summation-by-Parts (SBP) in
aerospace industry is impeded by their instability when applied to nonlinear problems [1]. The stability of ener-
gy-stable DG and SBP schemes are guaranteed for linear problems, but not for nonlinear problems [2]. Entropy-sta-
ble DG and SBP schemes are provably stable for nonlinear problems, but they are more costly [2]. Further, the
stability of entropy-stable DG is guaranteed only under exact integration [2, 3]. It is thus of interest to assess the
performance of these high-order schemes in practice and to determine which stability properties are important for
which classes of problems. In our work, an existing computer program is modified to simulate problems with
increasing levels of difficulty, using the following schemes: i) energy-stable DG ii) entropy-stable DG, iii) ener-
gy-stable SBP, and iv) entropy-stable SBP. For each problem, we study the accuracy of each scheme relative to their
cost and determine the maximum Courant-Friedrichs-Lewy condition for which each scheme becomes unstable. We
have implemented energy-stable SBP and DG schemes in one dimension and have verified that energy-stability
holds for the linear advection equation. It is expected that only the entropy-stable SBP scheme will be stable for the
most difficult problems. Our study will show the practical performance of high-order schemes and can serve as a
foundation for choosing schemes that meet the cost and stability demands of particular problems.

[1] Z. J. Wang et al., "High-order CFD methods: current status and perspective," International Journal for
Numerical Methods in Fluids, vol. 72, pp. 811-845, 2013.

[2] S. Shadpey and D.W. Zingg, "Energy- and Entropy-Stable Multidimensional Summation-by-Parts Discreti-
zations on Non-Conforming Grids," presented at the AIAA AVIATION Forum, 2019.

[3] J. Crean, J.E. Hicken, D.C. Del Rey Fernandez, D.W. Zingg, and M.H. Carpenter, "Entropy-stable summa-
tion-by-parts discretization of the Euler equations on general curved elements," Journal of Computational Physics,
vol. 356, pp. 410-438, 2018.

UnERD
2O19 • 36
AUG 13
H: Engineering Education
H1: Gendered Patterns in Senior Engineers’ Leadership Learning
Emily Macdonald-Roach, Dr. Cindy Rottmann, Dr. Andrea Chan, Prof. Emily Moore
1
Troost Institute for Leadership Education in Engineering, University of Toronto

Emphasis on engineering leadership education has become more prevalent in the last two decades, motivating
research and development of formal programs to teach leadership skills. However, research in this area has failed to
address situated leadership learning, that is how engineers learn to be leaders throughout their careers through
day-to-day experiences [1].
The Engineering Leadership Project addresses this gap by analyzing 29 interviews with senior engineering leaders
to see how they’ve learned leadership throughout their careers. This work looks at a subset of 8 interviews (1 man
and 1 woman from each of 4 different organizations) to compare the leadership learning of men and women in paral-
lel career paths. It aims to account for how leadership learning experiences may differ by gender, and identify differ-
ences in opportunities afforded to each group.

Inductive coding was used to methodically identify themes in the transcripts, and consolidate these to identify
patterns by gender within each organization and across the entire sample. Explicit mentions of gender were investi-
gated, in addition to implicit patterns by gender within mobility transitions, proud moments, and struggles. Results
will present key themes and highlight differences between the experiences of men and women, and emphasize the
importance of looking beyond explicit differences by gender to the subtleties that affect attitudes and representation.

A gendered analysis of leadership paths can inform future research about leadership learning opportunities. This can
further help educators, researchers, and employers ensure equitable and inclusive practices to better support the
engineering leadership and career development of various groups.

[1] C. Rottmann, D. Reeve, S. Kovalchuk, M. Klassen, M. Majkovic and E. Moore, "“Counting Past Two:” Engi-
neers’ Leadership Learning Trajectories", in 126th American Society for Engineering Education Annual Conference
& Exposition, Tampa, Florida, 2019, pp. 1-2.

37 • UnERD
2O19
AUG 13
CONTENTS: ABSTRACTS - POSTER PRESENTATIONS
A: CHEMICAL .....................................................................................39
B: BIOMATERIALS & BIOMEDICAL ...............................................42
C: BIOENGINEERING & BIOELECTRICAL ....................................54
D: MATERIALS SCIENCE ..................................................................57
E: ELECTRICAL & COMPUTER .......................................................61
F: MECHANICAL & INDUSTRIAL ...................................................65
G: CIVIL/MINERAL ............................................................................72
H: FOOD AND NUTRITION ...............................................................74

• 38
Abstracts – Poster Presentations
A: Chemical
A1: The Oxidative Potential of Particulate Matter from Transportation Sources
Sabrina Cupryk1, Cheol-Heon Jeong1, & Greg J. Evans1
1
Southern Ontario Centre for Atmospheric Aerosol Research (SOCAAR), University of Toronto

Particulate matter (PM) in air has been shown to cause oxidative stress leading to adverse health effects, such as lung
cancer and increased acute respiratory morbidity [1]. Oxidative potential (OP) is used as a biologically relevant
measurement for studying the health effects of PM [2].

A significant fraction of exposure to PM occurs during commuting. One area requiring further investigation is air
quality in subways and trains, due to the large variation in results seen between different systems [3]. Furthermore,
research is needed to investigate non-exhaust sources from automobiles, such as brake and tire dust, as existing stud-
ies have not distinguished these sources due to their interactions [4].

To compare the toxicity of these emission sources, PM samples were collected from the Toronto subway, the GO
train and automobile disc brakes. The metal content of the samples is determined through X-ray analysis. The deple-
tion of two key antioxidants in the lungs, ascorbate acid and glutathione, is simulated using OP assays. The obtained
results are analysed in order to compare the composition and toxicity of exposure to PM from the different sources.

It is expected the brake dust will have higher OP compared to subway and train PM due to their high transition metal
content. Variation in OP is also anticipated between the old and new subway lines. These findings provide localized
information on the toxicity of transportation sources in Toronto. Studying OP can better educate commuters, influ-
ence policy changes and improve transportation design to reduce exposure to toxic air pollutants.

[1] S. Weichenthal et al., “Oxidative burden of fine particulate air pollution and risk of cause-specific mortality
in the Canadian Census Health and Environment Cohort (CanCHEC),” Environ. Res., vol. 146, pp. 92–99, 2016.

[2] P. J. A. Borm, F. Kelly, N. Künzli, R. P. F. Schins, and K. Donaldson, “Oxidant generation by particulate
matter: From biologically effective dose to a promising, novel metric,” Occup. Environ. Med., vol. 64, no. 2, pp.
73–74, 2007.

[3] T. Moreno et al., “Oxidative potential of subway PM2.5,” Atmos. Environ., vol. 148, pp. 230–238, 2017.

[4] A. Thorpe and R. M. Harrison, “Sources and properties of non-exhaust particulate matter from road traffic:
A review,” Sci. Total Environ., vol. 400, no. 1–3, pp. 270–282, 2008.

39 • 2O19 UnERD
AUG 13
A2: Determining the Optimal Catalyst for Treatment of Post-smouldering Exhaust
Jasmeen Parmar1, Ashraf Amin1,2, Yu-Ling Cheng1,2
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Centre for Global Engineering, University of
Toronto;

Urban populations in developing countries suffer from public health issues due to lack of basic sanitation [1]. The
toilet research project at the Centre for Global Engineering are reinventing the toilet concept to provide a safe and
accessible waste treatment system for rural areas. The design consists of solid/liquid treatment subsystems to handle
fecal matter. Fecal matter is treated in a smouldering reactor, which is a slow, flameless, low temperature combus-
tion process [2]. The smouldering exhaust gas stream must be treated before it is released into the atmosphere. This
study aims to identify an optimal catalyst to oxidize CO and VOCs in smouldering exhaust. Catalyst selection is
based on high catalytic activity, high selectivity towards CO and VOC oxidation, resistance to deactivation, and low
cost. The optimal catalyst must also achieve a set of performance goals. Catalytic treatment will minimize emissions
to levels stipulated by ISO-30500 standards [3]. The catalyst should have an operating life of one year in the pres-
ence of deactivation agents like SOx and water vapor. A test unit was assembled for determining catalytic activity,
which consists of a gas feeding system (gas mixtures of known composition from cylinders) leading to a furnace/re-
actor assembly which encloses the catalyst. The composition of the post-catalytic gas stream is subsequently
analyzed to assess levels of oxidation using GC. The activity of different commercial catalysts is being tested in
different reaction conditions, such as the presence of SO2 and water, in order to replicate actual smouldering condi-
tions.

[1] A.Y. Katukiza, M. Ronteltap, C.B. Niwagaba, J.W.A Foppen, F. Kansiime, and P.N.L
Lens, “Sustainable sanitation technology options for urban slums,” Biotechnology
Advances, vol. 30, no. 5, pp. 964–978, Sep. 2012.

[2] G. Rein, “Smouldering Combustion Phenomena in Science and Technology”,

International Review of Chemical Engineering, Vol 1, pp 3-18, Jan 2009.

[3] Non-sewered sanitation systems - Prefabricated integrated treatment units - General

Safety and performance requirements for design and testing, ISO Standard 30500:2018, 2018.

UnERD
2O19 • 40
AUG 13
A3:The Oxidative Potential of Cooking Emissions
Loay Al Rashdi, Manpreet Takhar, Shunyao Wang, Qimei Wang, Arthur Chan
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto

Particulate matter (PM) from urban areas can cause oxidative stress and is associated with adverse health effects. [1]
Cooking emissions represent one of the most important sources of organic PM in urban areas and have been shown
to contribute 17-30% of total organic aerosol. [2] Despite its importance, the chemical composition of fresh and aged
PM from food cooking emissions and its oxidative potential (OP, or ability to cause oxidative stress) are poorly
understood. In addition, reactive oxygen species (ROS) play a key role in causing oxidative stress. ROS is formed
by the reduction of oxygen, consists of superoxide, OH radicals, and hydrogen peroxide. In this work, we use cook-
ing oils (canola, olive, peanut) as a surrogate to investigate the oxidative potential from direct (unoxidized) and indi-
rect (oxidized) exposure to cooking emissions using dithiothreitol (DTT) assay. Our preliminary results suggest that
direct exposure to cooking emissions correspond to low OP (5 - 20) irrespective of the cooking conditions. The OP
results from different oxidation pathways of cooking emissions in the atmosphere will be discussed in further detail.
Furthermore, the speciated composition of cooking emissions under conditions of exposure will be linked to its OP.

[1] D. W. Dockery and C. A. Pope, “Acute Respiratory Effects of Particulate Air Pollution,” Annual Review of
Public Health, vol. 15, no. 1, pp. 107–132, 1994.

[2] N. Gysel, P. Dixit, D. A. Schmitz, G. Engling, A. K. Cho, D. R. Cocker, and G. Karavalakis, “Chemical
speciation, including polycyclic aromatic hydrocarbons (PAHs), and toxicity of particles emitted from meat cooking
operations,” Science of The Total Environment, vol. 633, pp. 1429–1436, 2018.

A4: Synthesis of Dechlorinated BSubPc Precursors

Benjamin Shore

Boronsubphthalocyanines (BSubPcs) are being investigated by the Bender Laboratory for use in organic photovolta-
ics (OPVs) and organic light emitting diodes. BSubPcs are made by reacting phthalonitriles (precursor) with boron
trichloride. The phthalonitriles can have different amounts and positions of halides on the aromatic ring. It has been
shown that by making the BSubPc from precursor with 4 chlorines, the efficiency of the OPVs improved, but the
maximum current was reduced. It is theorized that phthalonitriles with fewer chlorines would have both increased
efficiency without reducing current, but phthalonitriles with different amounts of peripheral chlorines are not com-
mercially available. The goal of this project is to synthesize tri-chlorophthalonitrile, and 3,6-dichlorophthalonitrile
using purchasable materials. Tetrachlorophthalic anhydride has been shown to undergo selective dechlorination via
a hydro-dechlorination reaction, giving the tri-chloro and di-chloro phthalic acids. These phthalic acids can be
converted into phthalimides, then phthalic bis-amides, and finally phthalonitriles.

In the past two months, the mono-hydro-dechlorination reaction has been optimized to form the tri-chloro acid at
97% purity, with >75% yield. The imide reaction has been shown to be a purification step, with the imide product
being >99% pure. Later this summer, the reaction sequence to convert the imide to the phthalonitrile will be
performed, as well as further optimizing the bis-dechlorination pathway. The phthalonitriles made will allow the
graduate students in the laboratory to analyze the effects of these specific peripheral chlorines in BSubPcs, hopefully
forming OPVS with both increased efficiency and current.

41 • UnERD
2O19
AUG 13
 B: Biomaterials + Biomedical
B1: Processing Hand-object Interaction Detections from Egocentric Video of Individuals with
Spinal Cord Injury
Shannon Cavanagh1, 2, José Zariffa2, 3, 4, 5
1
Department of Mechanical Engineering, University of Alberta; 2KITE, Toronto Rehab, University Health Network; 3Institute of Biomate-
rials and Biomedical Engineering, University of Toronto; 4Rehabilitation Sciences Institute, University of Toronto; 5Edward S. Rogers. Sr.
Department of Electrical and Computer Engineering, University of Toronto

Hand function is an important determinant of independence for individuals with spinal cord injury (SCI). There is
currently a gap in the ability to directly quantify functional hand use in the home. In previous work, we demonstrated
the efficacy of applying machine learning algorithms to monitor functional hand use using egocentric videos of
patients with SCI [1]. To evaluate the algorithm’s ability to detect hand-object interactions, frame-by-frame results
must be converted into interpretable metrics. The current work is an optimization study of the processing methods.
Egocentric video was recorded from three SCI participants performing 38 activities of daily living in a home simula-
tion lab. The output of the hand-object interaction algorithm gives binary frame-by-frame results classifying interac-
tion. The results were filtered, normalized, and subjected to a threshold defining the limit between interaction and
non-interaction. The parameters varied were the length of filter, normalization method, and threshold value. The
area under the curve (AUC) on a receiver operating characteristic curve was used to characterize the ability of each
combination of parameters to classify hand-object interaction.

The test case with the greatest AUC was 0.86 with a threshold of 0.75, and max-min normalization applied. The
point with the optimal balance of true positive rate and false positive rate corresponded to 60 frames in the moving
average filter. Although the greatest AUC could be determined, there was noticeable variation in results between
datasets. Results from only three participants were assessed, so more datasets should be evaluated to verify the opti-
mal processing parameters.

[1] J. Likitlersuang, E. R. Sumitro, T. Cao, R. J. Visée, S. Kalsi-Ryan, and J. Zariffa, “Egocentric video: a new
tool for capturing hand use of individuals with spinal cord injury at home,” Journal of NeuroEngineering and Reha-
bilitation, vol. 16, no. 1, 2019.

UnERD
2O19 • 42
AUG 13
B2: Testing the Measures of Dynamic Postural Stability During Walking
Gagnon E1, Lee JW1,2, Yoo JE1,2, Shinya M3, Milosevic M4, Yokoyama H2, Masani K1,2

Institute for Biomaterials and Biomedical Engineering, University of Toronto; 2KITE - Toronto Rehabilitation Institute, University
1

Health Network; 3School of Integrated Arts and Sciences Hiroshima University; 4Department of Mechanical Science and Bioengineering,
Osaka University

Introduction
Elderly individuals or those with neurological disorders often display atypical walking characteristics, which can
increase the risk of falls. Accurate measures of dynamic postural stability, that is the ability to maintain balance
during a dynamic state, can help to evaluate their risk of falling. There are multiple measures proposed to assess the
dynamic postural stability during walking, including gait variability, local dynamic stability, and margin of stability.
Additionally, we have proposed to use the desired centre-of-pressure (dCOP), a virtual point on the ground resulting
in a moment equal to zero about the body’s centre-of-mass (COM). The misalignment between the dCOP and the
measured COP (dCOP-mCOP) can be used as a measure for dynamic postural stability. Here we tested the above
mentioned measures of dynamic postural stability during treadmill walking, to determine whether it is possible to
distinguish walking conditions with different levels of dynamic stabilities.

Methods
Fourteen able-bodied young males were asked to walk on a treadmill 1) naturally, 2) with a metronome, 3) with arm
restrictions, and 4) while completing a Stroop test. Body kinematics were measured using a motion capture system.
The body kinetics were measured using force plates located on the floor of the treadmill. Using the measured kine-
matics and kinetics, we calculated the aforementioned measures, and compared them with the four different walking
conditions.

Results and Discussion

The analyses are currently ongoing. The study outcome can contribute in the development of accurate assessment
protocol of the dynamic postural stability during walking, which can help those with atypical walking characteris-
tics.

43 • UnERD
2O19
AUG 13
B3: Pro-angiogenic protein delivery from microspheres for management of critical limb ischemia
(CLI)
Trevor Ung1,2, Eric Tawagi1,2, Paul Santerre1,2,3

Institute of Biomaterials and Biomedical Engineering, University of Toronto; 2Translational Biology and Engineering Program, Ted
1

Rogers Centre for Heart Research, University of Toronto; 3Faculty of Dentistry, University of Toronto

Critical limb ischemia (CLI) is characterized by arterial occlusion in the lower extremities, resulting in ischemia rest
pain, ulcers, or gangrene. While surgical revascularization (e.g. vascular bypass, angioplasty) is the preferred treat-
ment, not all patients are eligible, leading to high rates of amputation and limb loss, cardiovascular comorbidities,
and mortality. Minimally invasive alternatives for lower limb revascularization may help improve management and
outcomes of CLI. Angiogenic growth factors help promote new blood vessel growth into ischemic tissues. Biore-
sorbable polymeric microspheres are frequently used to provide controlled, sustained delivery of growth factors, via
localized, minimally invasive injections at the diseased site. However, currently used polymeric vehicles are pro-in-
flammatory, and unsuitable for inflammatory diseases like CLI. Recently, low-inflammatory, biodegradable poly-
urethane microspheres were developed and evaluated for use as a protein delivery system. Polyurethane micro-
spheres are expected to efficiently encapsulate growth factors and show a controlled release profile, while maintain-
ing growth factor activity. Microspheres were prepared using a method developed in our lab that provided low size
dispersity. Microspheres were assessed for their ability to encapsulate and release model proteins and growth
factors. Released growth factor activity was assessed by characterizing their proliferative effect on endothelial cells.
Microspheres released model protein in a controlled manner. The activity of growth factors was confirmed with a
cell proliferation assay. Current work includes testing different stabilizers to maintain growth factor activity, and
tuning the release kinetics of the protein. Ultimately, the development of microsphere vehicles for angiogenic
growth factors may allow for the management of CLI.

B4: Optimization of conditions to isolate and grow cells to engineer living replacement heart valves
Monica Lecce1,2, Neda Latifi2,3, Craig A. Simmons2,3,4
1
Department of Chemical Engineering, University of Toronto; 2Translational Biology and Engineering Program, Ted Rogers Centre for
Heart Research; 3Department of Mechanical and Industrial Engineering, University of Toronto; 4Institute of Biomaterials and Biomedi-
cal Engineering, University of Toronto

Congenital heart disease affects ~1% of Canadian newborns, often involving defective heart valves [1]. Prosthetic
replacement valves cannot grow with the child and repair materials have limited life-spans, necessitating multiple
surgeries early in life. Tissue engineering offers the unique opportunity to produce growable living heart valves that
last a lifetime. Heart valve engineering requires cells to produce valve-like tissue. We previously found that human
umbilical cord cells, a personalized source of cells, are readily expanded in culture and produce extracellular matrix
proteins found in native valve tissue. The next step in developing engineered heart valves requires testing in pigs.
This study aims to optimize the culture conditions to isolate and grow pig umbilical cord cells. Cells were counted
over time in one of four culture media: complete RoosterBio, complete StemMACs, basal StemMACs with 20%
fetal bovine serum (FBS) and Dulbecco’s Modified Eagle Medium (DMEM) with 20% FBS. Preliminary results
showed that more than 50% of cells in RoosterBio and StemMACs complete media were detached after 48 hours.
Conversely, cells in DMEM+FBS and StemMACS+FBS for 7 days increased in number by 100% and 70%, respec-
tively. Future experiments will determine the effects of seeding density and time in culture on cell growth and extra-
cellular matrix synthesis using DMEM+20% FBS, determined to be the optimal medium. The results of this study
provide a strong foundation towards testing of engineered heart valve tissue in a pig model.

[1] Public Health Agency of Canada (2013) Congenital Anomalies in Canada 2013: A Perinatal Health Surveillance
Report. p33-39

UnERD
2O19 • 44
AUG 13
B5: Suppression of fibrosis in human cardiac fibroblasts in vitro with sustained release of C-type
natriuretic peptide from nanoparticles
Shirley Chung1,2,3, Nataly M. Siqueira2,3, Craig A.G. Simmons2,3,4, J. Paul Santerre2,3,5

Department of Biology, University of Waterloo, 200 University Ave W, Waterloo, ON N2L 3G1, Canada; 2Ted Rogers Centre for Heart
1

Research, 661 University Avenue, 14th Floor, Toronto, ON M5G 1M1, Canada; 3Institute of Biomaterials & Biomedical Engineering,
University of Toronto, 164 College Street, Toronto, ON M5S 3G9, Canada; 4Department of Mechanical and Industrial Engineering,
Faculty of Applied Science and Engineering, Toronto, Ontario M5S 3G8, Canada; 5Faculty of Dentistry, University of Toronto, Ontario
M5G 1G6, Canada

When the myocardium becomes damaged through events such as a myocardial infarction, a fibrotic scar forms
resulting in reduced ventricle contractility, decreased ability for the heart to pump blood throughout the body, and
increased susceptibility for the patient to undergo heart failure. [1, 2] In a healthy heart, cells naturally secrete C-type
natriuretic peptide (CNP), an anti-fibrotic cytokine that has strong effects on cardiac fibroblast function, fibrosis,
and ECM deposition. [3, 4] CNP functions by antagonizing the well-known transforming growth factor-β1 (TGF-β
1) pathway, thus inhibiting TGF-β1’s role to induce fibrosis and therefore reducing differentiation to myofibroblasts.
[5–9] We hypothesize that the sustained release of exogenous CNP from nanoparticles will reduce fibrosis in dam-
aged myocardium and suppress differentiation of human cardiac fibroblasts to myofibroblasts. Human cardiac fibro-
blasts will be cultured in collagen gels and treated with TGF-β1, CNP, TGF-β1 alongside CNP, or non-treated as a
control. TGF-β1 will be used to induce cell contraction and CNP will be delivered alone or through controlled
release from degradable-polar-hydrophobic-ionic polyurethane (D-PHI) nanoparticles that will either be uncoated
or coated with heparin/poly-lysine. Following treatment, the contractility of the cells will be monitored through
measurement of the surface area of the gels. We expect TGF-B1 to induce gel contraction (as indicated by a smaller
gel area), which will be reversed by treatment with CNP. Results obtained from this project will aid in determining
a dose-dependent range of CNP for use as an anti-fibrotic agent to counter fibrosis that arises due to a myocardial
infarction.

[1] R. J. Richardson, “Parallels between vertebrate cardiac and cutaneous wound healing and regeneration,”
Nat. Regen. Med., vol. 3, no. 21, 2018.

[2] V. Talman and H. Ruskoaho, “Cardiac fibrosis in myocardial infarction — from repair and remodeling to
regeneration,” Cell Tissue Res., pp. 563–581, 2016.

[3] H. J. Jansen and R. A. Rose, “Natriuretic Peptides: Critical Regulators of Cardiac Fibroblasts and the Extra-
cellular Matrix in the Heart,” in Advances in Biochemistry in .

45 • UnERD
2O19
AUG 13
B6: An Image-Based Approach to Study the Impact of Local Neighborhood Makeup on Cell Com-
petition
Lauren Banh1, Michael Paraskevopoulos1, Alison P. McGuigan1,2

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Institute of Biomaterials and Biomedical
1

Engineering, University of Toronto

Cell competition is the phenomenon whereby cells of greater fitness eliminate less fit neighboring cells [1]. This
phenomenon has been implicated in several contexts, such as cancer and Alzheimer's disease [2], [3]. Therefore,
understanding the factors that shape cell competition could be informative in the efforts against these diseases.
Recent work has demonstrated that cell competition mediated by fitness-sensing can be contact-dependent [4], how-
ever it is unknown how its outcome is affected by the spatial distribution of competing cells. Thus, the aims of the
project presented are: (i) to identify a method for analyzing the spatial distribution of cells grown on 2D surfaces,
and (ii) to apply that method to decipher the interactions between competing cancer cells. For this reason, we adopt-
ed a previously published approach to analyze the spatial distribution of competing cells from fluorescent microsco-
py images [5]. The algorithm was implemented in MATLAB® where the fluorescent images are segmented to
construct the Voronoi diagram. The diagram depicts the boundaries of each cell, identifies the cells from each com-
peting population, and determines if the cell is eliminated. This information is combined to compute the likelihood
of cell elimination in relation to the cell’s neighborhood composition. Preliminary results have indicated that the
algorithm is a good approximation for cells with polygonal morphology but performs poorly if that condition is not
met. Therefore, further development is needed prior to using this method in analyzing the interactions between the
cancer cells that were previously shown to compete by our lab.

[1] D. Grifoni and P. Bellosta, “Drosophila Myc: A master regulator of cellular performance,” Biochimica et
Biophysica Acta - Gene Regulatory Mechanisms, vol. 1849, no. 5. Elsevier, pp. 570–581, 01-May-2015.

[2] D. S. Coelho and E. Moreno, “Emerging links between cell competition and Alzheimer’s disease,” J. Cell
Sci., vol. 132, no. 13, p. jcs231258, Jul. 2019.

[3] S. J. E. Suijkerbuijk, G. Kolahgar, I. Kucinski, and E. Piddini, “Cell competition drives the growth of intesti-
nal adenomas in Drosophila,” Curr. Biol., vol. 26, no. 4, pp. 428–438, Feb. 2016.

[4] E. Madan et al., “Flower isoforms promote competitive growth in cancer,” Nature, p. 1, Jul. 2019.

[5] A. Bove, D. Gradeci, Y. Fujita, S. Banerjee, G. Charras, and A. R. Lowe, “Local cellular neighborhood
controls proliferation in cell competition,” Mol. Biol. Cell, vol. 28, no. 23, pp. 3215–3228, Nov. 2017.

UnERD
2O19 • 46
AUG 13
B7: Identifying Brain State Transitions In Epileptic Seizures Using Feature Selection
Nizar Islah1,2, Gerard O’Leary1,3, Roman Genov3, Taufik A. Valiante1,4,5

Krembil Research Institute, University Health Network; 2Dept. of Computer Science, University of Toronto; 3Dept. of Electrical and
1

Computer Engineering, University of Toronto; 4Institute of Biomaterials and Biomedical Engineering, University of Toronto; 5Division of
Neurosurgery, Dept. of Surgery, University of Toronto

Epilepsy is a neurological disorder which affects over 65 million people worldwide. A major challenge in the study
of this condition is understanding how brain activity changes in the period before and after a seizure (peri-ictal), as
there is a lack of robust methods to reliably identify the spatiotemporal brain-state changes in this period. The state
transition sequences can vary greatly from patient to patient as each brain and electrode positioning is unique.
Recent work has shown that distinct state transitions occur in the minutes leading up to and during a seizure [1]. This
work presents a new method of identifying peri-ictal state transitions.

First, signal energy features in physiological signal bands are extracted from intracranial EEG (iEEG) data. A feature
selection algorithm is used to quantify the change in important features over time. This is achieved by comparing
windowed peri-ictal data to a reference window of random samples from the patient’s iEEG data. We then use
change-point detection to segment the windows into separate states. K-means clustering on the signal energy
features groups together similar states, creating a unique set of distinct states.

Our findings show that a unique set of brain states can be consistently identified and show similarities across
seizures. This is an important step toward further understanding abnormal brain activity in epilepsy. In addition,
physicians could potentially have access to a live comprehensive overview of patients' peri-ictal brain activity,
which can identify regions of interest for surgical or device-based intervention, providing new opportunities for
clinical treatment.

[1] S. P. Burns, "Network dynamics of the brain and influence of the epileptic seizure onset zone", Proc. Nat.
Acad. Sci. USA, vol. 111, no. 49, pp. E5321-E5330, 2014.

47 • UnERD
2O19
AUG 13
B8: siRNA Mediated Knockdown of a Key Invasive Driver of Glioblastoma
Emma Jameson1, Ellie Arnold2, Elise Malek-Adamian3, Phuong U. Le4, Anika Meng4, Saúl Martínez-Montero3,
Kevin Petrecca5, Masad J.Damha3, and Molly Shoichet1,2,6

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Department of Chemistry, University of Toronto;
1

Department of Chemistry, McGill University; 4Division of Engineering Science, University of Toronto; 5Department of Neurology and
3

Neurosurgery, Montreal Neurological Institute and Hospital, McGill University; 6Institute of Biomaterials and Biomedical Engineering,
University of Toronto

Glioblastoma multiforme (GBM), a lethal form of brain cancer, presents significant treatment limitations given its
aggressive nature [1]. The challenges associated with treating this cancer are largely a result of the involvement of
glioblastoma stem cells (GSCs) which are invasive, treatment-resistant brain cancer cells hypothesized to cause
cancer recurrence [2]. Downregulated in renal cell carcinoma (DRR) has been identified as a genetic driver of GSC
self-renewal and invasion and presents itself as a potential target in the treatment of GBM [3]. The objective of this
project is to limit GSC invasion by downregulating DRR through the administration of modified DRR-specific
small interfering nucleic acids (siRNA). The siRNAs used in this study were chimeras of DNA and 2’-deoxy-2’-flu-
oro-beta-D-arabinose (FANA) which provides increased nuclease stability and mRNA affinity. The stability of the
different siRNAs was evaluated by incubating them with serum and quantifying the siRNA remaining using gel
electrophoresis. Knockdown of DRR by the different siRNAs was determined by transfecting glioblastoma cells and
evaluating the expression of DRR using quantitative PCR and western blot techniques. Our preliminary results have
confirmed the added stability of siRNAs incorporating FANA while showing similar levels of DRR knockdown in
all siRNA DNA/FANA variants. Future experiments will investigate the invasion of transfected glioblastoma cells
in a hyaluronan-based hydrogel [4] to determine the reduction in invasive behavior as a result of DRR-specific
siRNA knockdown. Through this study, we will further our understanding of glioblastoma invasion with the ultimate
goal of contributing to the development of more effective GBM therapies.

Acknowledgements
We are grateful to the Canadian Institute for Health Research and the Natural Sciences & Engineering Council of
Canada (CGSD to A.E.A. and USRA to E. J.) for their support in this project.

[1] E. C. Holland, "Glioblastoma multiforme: The terminator," Proceedings of the National Academy of Scienc-
es, vol. 97, no. 12, p. 6242, 2000.

[2] F. Orzan et al., "Genetic Evolution of Glioblastoma Stem-Like Cells From Primary to Recurrent Tumor,"
STEM CELLS, vol. 35, no. 11, pp. 2218-2228, 2017/11/01 2017.

[3] A. Dudley et al., "DRR regulates AKT activation to drive brain cancer invasion," Oncogene, Original Article
vol. 33, p. 4952, 10/21/online 2013.

[4] R. Y. Tam et al., "Rationally Designed 3D Hydrogels Model Invasive Lung Diseases Enabling High-Content
Drug Screening," Advanced Materials, vol. 31, no. 7, p. 1806214, 2019/02/01 2019.

UnERD
2O19 • 48
AUG 13
B9: Tuning the properties of engineered hydrogels to modulate macrophage phenotype
Danielle Karakas1, Laura C. Bahlmann1, Rob C. Laister2, Molly S. Shoichet1,3,4
1
Department of Chemical Engineering & Applied Chemistry, University of Toronto ; 2Cancer Institute, Princess Margaret Hospital ;
3
Institute of Biomaterials and Biomedical Engineering, University of Toronto ; 4Department of Chemistry, University of Toronto

Despite advances in treatments, there remains a significant subset of cancer patients that are unresponsive to treat-
ment and/or experience relapse. These relapses are often associated with macrophage infiltration; however, mecha-
nisms of polarization to a pro-tumour phenotype in the microenvironment are largely unknown. [1] Due to the
dynamic nature of in vivo models, it is difficult to isolate the factors that contribute to the phenotypic changes. The
purpose of this study is to develop an in vitro hydrogel model for the analysis of individual biochemical cues on
macrophage response. A cross-linked oxime hydrogel, synthesized with oxyamine-functionalized gelatin and alde-
hyde-functionalized hyaluronic acid (HA), constitutes the scaffold. The properties of such hydrogels can be tuned in
terms of protein additives, porosity and polymer content. THP1 cells, a human monocytic cell line which can be
synthetically differentiated into macrophages, was selected for the study. Following 24 hours of culture with the
hydrogels, cells were collected for RNA processing and quantitative polymerase chain reaction (qPCR) to quantify
gene expression to determine the impact of the hydrogel on the cells. Preliminary data showed that Transglutami-
nase 2 (TGM2) and Arginase 1 (ARG1), markers of anti-inflammatory, immunosuppressive response were elevated
in hydrogels containing collagen I and HA. In future studies it is expected that increased porosity and protein content
will shift cells towards a pro-tumour phenotype that will promote cell spreading and migration. These studies will
allow for the elucidation of criterion that impact macrophage-like polarization in hydrogels and will inform the
development of disease models for advanced-stage cancer.

[1]D. W. Scott and C. Steidl, “The classical Hodgkin lymphoma tumor microenvironment: macrophages and gene
expression-based modeling,” Hematology, vol. 2014, no. 1, pp. 144–150, Dec. 2014.

49 • UnERD
2O19
AUG 13
B10: Using digital microfluidics to implement an immunofluorescence pipeline on mouse neuronal
and glial cells in vitro
Alyssa Iglar1,2, Erica Scott1,3, Aaron Wheeler1,3

nstitute of Biomaterials and Biomedical Engineering, University of Toronto; 2Department of Mechanical and Industrial Engineering,
1

University of Toronto; 3Department of Chemistry, University of Toronto

We use immunofluorescent cell staining to visualize [1] and classify cells, with overall assessment of cell type
heterogeneity in a primary cell culture from the mouse subventricular zone. Currently, the in vitro staining protocol
is carried out using a pipette to manually dispense the necessary reagents. However, this technique relies on the use
of larger reagent volumes and increases the likelihood of culture dehydration, which results in lower quality images.
We have designed an assay using digital microfluidics (DMF) to implement the staining protocol and attempt to
address these limitations. The two-plate DMF devices consist of an indium tin oxide (ITO)-coated top plate, which
houses hydrophilic sites permitting cell culture adherence, and a bottom plate consisting of electrodes used to
manipulate microliter volumes of liquid [2]. Applying a voltage across the device actuates the electrodes and facili-
tates droplet movement, allowing the transfer of reagents over the hydrophilic sites [2]. Our DMF pipeline uses
lower reagent volumes, avoids sample dehydration, and has the potential to increase the consistency of the staining
protocol due to less reliance on technical skill with a pipette.

Currently, we have demonstrated that the DMF platform is capable of performing this immunofluorescent assay.
However, we have yet to see a significant difference in terms of image quality between the two staining methods.
These assessments will validate whether the DMF immunofluorescent assay can be an alternative to conventional
pipette-based staining protocols, while increasing our understanding of cell heterogeneity associated with mouse
subventricular cell cultures.

[1] M. Bruckner, "Microscopy", Microbial Life: Educational Resources, 2019. [Online]. Available: https://serc.-
carleton.edu/microbelife/research_methods/microscopy/index.html. [Accessed: 27- Jul- 2019].

[2] Ng, A., Chamberlain, M., Situ, H., Lee, V. and Wheeler, A. (2015). Digital microfluidic immunocytochemis-
try in single cells. Nature Communications, [online] 6(1). Available at: http://microfluidics.utoronto.ca/papers/n-
comms8513.pdf [Accessed 27 Jul. 2019].

UnERD
2O19 • 50
AUG 13
B11: Generation of retinal organoids from human pluripotent stem cells for photoreceptor trans-
plantation
Emily Tsao1, Margaret Ho1,2, Molly Shoichet1,2,3,4
1
Institute of Biomaterials and Bioengineering, University of Toronto; 2Donnelly Centre for Cellular & Biomolecular Research, University
of Toronto; 3Department of Chemical Engineering and Applied Chemistry, University of Toronto; 4Department of Chemistry, University
of Toronto

Retinal cell death is a principal cause of permanent visual impairment, owing to the retina’s inability to regenerate.
Delivery of healthy photoreceptors to the injured retina via an injectable hyaluronan methylcellulose (HAMC)
hydrogel has potential to repair damaged tissue and restore vision [1]. However, high-yield isolation of
patient-compatible human photoreceptors is severely limited by supply and poor survival in vitro [2]. Stem
cell-derived retinal organoids offer a promising source of photoreceptors that can be personalized to patients for
transplantation. Furthermore, these organoids can be used to model retinal degeneration in vitro and adapted for
drug screening platforms. This study aims to characterize photoreceptor development throughout retinal organoid
differentiation to identify the optimal stage for cell harvest and transplantation.

We have generated retinal organoids from both human embryonic stem cell and induced pluripotent stem cell lines.
Neuroretinal vesicles were manually excised and further matured in suspension culture supplemented with taurine
and retinoic acid [3]. Organoids were collected at various time points for immunohistochemistry and gene expres-
sion analysis. Preliminary data from staining and quantitative RT-PCR have demonstrated that these organoids
express early progenitor genes (e.g. orthodenticle homeobox 2), and at later time points, photoreceptor-specific
genes (ex. rhodopsin, recoverin, s-opsin). Future work will explore survival of photoreceptors at various stages of
maturity encapsulated in the HAMC hydrogel. The results from this study will link stages of organoid development
to photoreceptor differentiation and demonstrate the feasibility of retinal organoids as a source of patient-derived
photoreceptors for retinal regeneration.

[1] B. G. Ballios, M. J. Cooke, D. van der Kooy, and M. S. Shoichet, “A hydrogel-based stem cell delivery
system to treat retinal degenerative diseases,” Biomaterials, vol. 31, no. 9, pp. 2555–2564, Dec. 2010.

[2] S. Llonch, M. Carido, and M. Ader, “Organoid technology for retinal repair,” Developmental Biology, vol.
433, no. 2, pp. 132-143, Jan. 2018.

[3] A. Gonzalez-Cordero, K. Kruczek, A. Naeem, M. Fernando, M. Kloc, J. Ribeiro, D. Goh, Y. Duran, S. J. I.

Blackford, L. Abelleira-Hervas, R. D. Sampson, I. O. Shum, M. J. Branch, P. J. Gardener, J. C. Sowden, J. W. B.
Bainbridge, A. J. Smith, E. L. West, R. A. Pearson, and R. R. Ali, “Recapitulation of Human Retinal Development
from Human Pluripotent Stem Cells Generates Transplantable Populations of Cone Photoreceptors,” Stem Cell
Reports, vol. 9, no. 3, pp. 820-837, Aug. 2017.

51 • UnERD
2O19
AUG 13
B12: Investigating invasion-related genetic biomarkers in glioblastoma cells using a 3D-biomimetic
hydrogel platform
Allysia A. Chin1, Laura J. Smith1,2, Molly S. Shoichet1,2,3

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Institute of Biomaterials and Biomedical
1

Engineering, University of Toronto; 3Department of Chemistry, University of Toronto

Glioblastoma (GBM) is the most common and deadliest brain cancer worldwide. Patients have few therapeutic
options because GBM is highly invasive, making surgical resection inadequate [1]. Moreover, conventional 2D
tissue-culture plastic-based drug screening platforms are unable to assess the invasive behaviour of cancer cells
because they lack the relevant physical and biochemical microenvironmental complexity which orchestrate cell
invasion in vivo. Thus, there is a need for 3D-biomimetic cell invasion models to effectively screen anti-invasive
chemotherapeutics, as well as study the properties of aggressive and invasive cell populations. Herein, we hypothe-
size that a 3D hydrogel model can be utilized to recapitulate the invasive behaviour of GBM cells in vivo, and study
the properties of aggressive and invading cell populations. Hydrogels were prepared using click chemistry, incorpo-
rating hyaluronan, methylcellulose, degradable and adhesive peptides to model the GBM niche in vitro [2]. When
seeded on top of these hydrogels, a small subset of patient-derived GBM cells invaded into the material while anoth-
er subset of cells proliferated on top of the gel and were non-invading. We were able to isolate live invading and
non-invading GBM cell populations by using an enzymatic degradation strategy. Preliminary studies have shown
differences in proliferative capacity and invasion-related gene expression between invading and non-invading popu-
lations. Ongoing experiments seek to generate gene expression profiles for each population using RNA sequencing
and perform high-throughput drug screens to test the efficacy of anti-invasive drugs. This work demonstrates a
potential platform for probing and eventually targeting mechanisms involved in cellular invasion in GBM.

[1] M. Ammirati, N. Vick, Y. Liao, I. Ciric, and M. Mikhael, “Effect of the Extent of Surgical Resection on
Survival and Quality of Life in Patients with Supratentorial Glioblastomas and Anaplastic Astrocytomas,” Neuro-
surgery, vol. 21, no. 2, pp. 201–206, Aug. 1987.

[2] R. Y. Tam et al., “Rationally Designed 3D Hydrogels Model Invasive Lung Diseases Enabling High-Content
Drug Screening,” Adv. Mater., vol. 31, no. 7, p. 1806214, Feb. 2019.

We also acknowledge the funding received from the Amgen Foundation, NSERC and CIHR.

UnERD
2O19 • 52
AUG 13
B13: Encapsulation-free Controlled Release of Pituitary Adenylate Cyclase Activating Polypeptide
from a Biodegradable Hydrogel-Nanoparticle Composite
Kevin Da1, Eric Ho1, Molly Shoichet1,2,3,4

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Institute of Biomaterials and Biomedical
1

Engineering, University of Toronto; 3Department of Chemistry, University of Toronto; 4Donnelly Centre for Cellular & Biomolecular
Research, University of Toronto

Controlled therapeutic release is vital to overcome off-target effects and the frequency of drug administration.
Poly-(lactic-co-glycolic)-acid (PLGA) remains a popular polymer for nanoparticle drug and protein delivery due to
its biocompatibility, biodegradability, and FDA approval [1]. Typically, PLGA is utilized to encapsulate proteins
within nanoparticle or microparticle formulations; however, poor encapsulation efficiency and low protein loading
limit the efficacy of this approach. To overcome this, our research group has explored controlling the release of ther-
apeutics by electrostatically adsorbing positively-charged proteins onto the negatively-charged PLGA nanoparticles
dispersed in a hyaluronan-methylcellulose (HAMC) hydrogel [2].

The objective of this project was to identify parameters that affect nanoparticle degradation rate, and thus release
rate of entrapped therapeutics. We explored the release of the model therapeutic pituitary adenylate cyclase activat-
ing polypeptide (PACAP), a positively charged peptide. PLGA nanoparticles were synthesized with varying proper-
ties with a water/oil/water double emulsion procedure to modulate the diameter and surface charge of the particles.
The nanoparticles were dispersed in HAMC and soluble PACAP was adsorbed to the particles therein. PACAP
concentration in the release samples will be assessed with high pressure liquid chromatography-tandem mass spec-
trometry. We expect that the nanoparticles formulated with greater molecular weights and smaller diameters will
extend PACAP release from the hydrogel-nanoparticle composite. These studies will enable us to design optimized
protein delivery vehicles for a mouse model of stroke injury for ultimate use in local, controlled drug delivery to the
central nervous system.

[1] H. K. Makadia and S. J. Siegel, “Poly Lactic-co-Glycolic Acid (PLGA) as Biodegradable Controlled Drug
Delivery Carrier,” Polymers, vol. 3, no. 3, pp. 1377–1397, 2011.

[2] M. M. Pakulska et al., “Encapsulation-free controlled release : Electrostatic adsorption eliminates the need
for protein encapsulation in PLGA nanoparticles,” Sci Adv., vol. 2, no. 5, 2016.

53 • UnERD
2O19
AUG 13
 C: Bioengineering & Bioelectrical
C1: Investigating the degradation of industrial lignin compounds by anaerobic microcosms using
Biochemical Methane Potential(BMP) assays
Chen Yu Wang, Anupama Sharan, Elizabeth Edwards and Emma Master
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto

Lignin is the second most abundant natural polymers in the world and makes up 20-30% of biomass. [1] Degrada-
tion of lignin plays an important role in improving the valorization of earth’s biomass resources as it is often the
rate-limiting step in biobased industrial production process. From literature evidence, some anaerobic acetogenic
bacteria can degrade lignin through a unique enzymatic system known as o-demethylation.[2][3] The enzyme
system has been tested with monolignols but not with complex lignin substrates. Therefore this project is intended
to identify the anaerobic microcosms that might produce these enzymes by testing their growth with industrial
lignin. In anaerobic degradation of biomass, the complex carbon compounds are degraded to methane and carbon
dioxide biogas.[4] Using this principle, a biochemical methane potential assay is set up to test microcosms sourced
from anaerobic pulp mill granule (AG). In different sets of 160ml serum bottles, the anaerobic microcosms are fed
with different industrial lignin and stored in 37 incubator. The biogas production is measured for each bottle with
syringe over time and used to calculate degradability of the different types of lignin. After 30 days of incubation,
degradability ranges from 52% for kraft lignin with vanillin to 4% for organosolv lignin for microcosms sourced
from AG. The evidence of degradation indicates the present of lignin-active enzyme in the microcosms. The
lignin-active enzymes will then be extracted and studied to potentially identify the genes in the microbial communi-
ties that catalyze the degradation; which will improve our current understanding of enzymatic lignin degrading
process.

[1]“What is Lignin? | Lignoworks”. [Online]. Available: http://www.icfar.ca/lignoworks/content/what-lignin.html.

[Accessed: 27-Jul.-2019].

[2]D. Naidu and S. Ragsdale W., “Characterization of a Three-Component Vanillate O-Demethylase from Moorella
thermoacetica”, 01-Jun.-2011.

[3]S. A, “The ether-cleaving methyltransferase system of the strict anaerobe Acetobacterium dehalogenans: analysis
and expression of the encoding genes”, Mar. 2009.

[4] M. Gould Charles, “Bioenergy and Anaerobic Digestion”, Nov. 2014.

UnERD
2O19 • 54
AUG 13
C2: Exploring the Mechanism of Coordinated Reset Stimulation for Long-Term Seizure Reduction
in Epilepsy
Bipasha Goyal1, Azadeh Naderian2,3, Gerard O’Leary2,3,4, Taufik Valiante2,3,5
1
Division of Engineering Science, University of Toronto; 2Neuron to Brain Laboratory; 3Krembil Research Institute; 4Institute of Biomate-
rials and Biomedical Engineering, University of Toronto; 5Division of Neurosurgery, Department of Surgery, University of Toronto

Epilepsy is a chronic disorder of the nervous system characterised by intense seizures due to abnormal synchronized
firing of neurons. For patients that do not respond to medication and cannot have seizure-generating regions
removed, electrical neurostimulation is promising. Neurostimulation can induce long-term changes in neuronal
connections. Coordinated reset (CR) is a novel strategy that uses specific stimulation patterns to desynchronise
neurons and thus prevent seizures. However, the mechanism realizing these effects is unknown. Our aim is to under-
stand how CR alters neural synchrony. We hypothesise changes in excitability and synchrony are interrelated.
Mouse hippocampal slices, used for simplicity, are stimulated and recorded from using multi-electrode arrays. After
taking baseline recordings, we induce long- term-potentiation (LTP) to increase neuronal excitability and administer
CR stimulation. The effectiveness of CR is tested by replacing it with long-term-depression (LTD) or no stimulation.
Changes in synchrony are studied by measuring the initial slope and amplitude of the neuronal response after each
step of the procedure. LTP is expected to increase both parameters. CR and LTD will have opposite effects, showing
weakened connections, and thus desynchronisation. Due to low frequency stimulation over short intervals, CR
reduces tissue damage and is thus superior to standard deep brain stimulation like LTD. No stimulation is the
control. Currently, baseline and post LTP recordings have been analysed. Results are as expected. Further, effects of
LTP remain for over 3 hours post induction. Next steps include showing the effects of CR. We will then explore stim-
ulation strength, frequency, pattern, and duration to find optimal parameters for depotentiating excitability which
can help devise potent treatments for epilepsy.

C3: Charge Balanced Arbitrary Waveform Generation for Neuromodulation using Embedded
Devices
Aditya Saigal1, Gerard O’Leary2,3,4, Roman Genov5,6,7
1
Division of Engineering Science, University of Toronto; 2Institute
5
of Biomaterials and Biomedical Engineering; 3Krembil Research
Institute; Neuron to Brain Lab; Electronics Group and Biomedical Engineering Group; 6Intelligent Sensory Microsystems Laboratory;
4 5

7
The Rogers Sr. Department of Electrical and Computer Engineering University of Toronto

Neural activity is often recorded in the form of electrical signals using microelectrode arrays. Occurrence of seizures
in patients with neural disorders like epilepsy can be determined by analyzing these signals. Currently, electrical
stimulators modulate neurons using square waves to reverse seizure-onset upon detection. It has been shown that
more complex, custom waveforms control seizures much more precisely. The purpose of this project is to create a
low-power, compact and low-latency Arbitrary Waveform Generator to be used on a device monitoring and stimu-
lating neurons in patients with epilepsy. We define latency as the number of clock-cycles that produce a single value.
A compact and power-efficient device will take minimal on-chip area and energy to produce waveforms. To mini-
mize tissue damage from charge accumulation, the waveform must be charge balanced, i.e. it should have no net
DC-current. This is done through Binary Exponential Charge Recovery. The generator is implemented by program-
ming an FPGA (an embedded device) for low-latency. This is connected to a processor that adjusts parameters like
waveform frequency and shape. Multiple waveform generation algorithms will be custom-built and tested while
designing. CORDIC and Look-Up-Table algorithms were tested and had similar performance. More algorithms such
as the Fast-Fourier-Transform will be developed and tested using the metrics and by observing the impact of stimu-
lating live tissue. The best solution will be selected. This device provides a robust way of digitally communicating
with neurons. Furthermore, the design objectives allow the module (after slight modifications) to be placed on
implantable devices for treating epilepsy and possibly other conditions.

55 • UnERD
2O19
AUG 13
C4: Reduced Model Approach Yields Necessary Conditions for Reliable Asynchronous Neural
Information Transmission
Sayan Faraz1, Milad Lankarany1, 2
1
Krembil Brain Institute, University Health Network; 2Institute of Biomaterials and Biomedical Engineering, University of Toronto

Human cognition relies on interactions between modular brain structures. This requires clusters of neurons to
reliably transmit information through ‘cables’ made of sequential layers of neurons connected by synapses. Each
neuron receives a sum of signals from neurons in the previous layer, weighted by synaptic weights. During transmis-
sion, neurons in a layer can either all fire at once, or spike asynchronously. Synchronous and asynchronous transmis-
sion use different representations of information [1]. Synchronous propagation is well-understood [2]. However,
most factors affecting asynchronous propagation remain unexplored due to problem complexity.
Theoreticians assume that all neurons in a fully-connected feed-forward network (fcFFN) are identical when explor-
ing propagation [1][3]. This assumption is used to reduce the representation of synaptic weights from an N*N matrix
to an N*1 vector. The reduced model vastly simplifies the problem while retaining the behaviour of the matrix repre-
sentation.

Least-squares optimization on the reduced model yields biologically feasible weights closely resembling weights
derived by applying gradient descent optimization on a full-model fcFFN. Reliable transmission of asynchronous
spikes in a reduced model is demonstrated using simulations and its behaviour is verified against a full model.
Preliminary results from reduced model simulations yield optimal values for network size (500) and background
synaptic noise amplitude (25pA) for asynchronous propagation. The effects of synaptic learning on transmission are
currently being investigated, and the reduced model is being extended to represent sparse neural networks. The
reduced model simplifies the search for biological conditions allowing for reliable communication in the brain,
advancing the understanding of human cognition.

[1] A. Kumar, S. Rotter, and A. Aertsen, “Spiking activity propagation in neuronal networks: reconciling differ-
ent perspectives on neural coding,” Nat Rev Neurosci, vol. 11, no. 9, pp. 615–627, Sep. 2010.

[2] M. Diesmann, M.-O. Gewaltig, and A. Aertsen, “Stable propagation of synchronous spiking in cortical
neural networks,” Nature, vol. 402, no. 6761, pp. 529–533, Dec. 1999.

[3] M. C. W. van Rossum, G. G. Turrigiano, and S. B. Nelson, “Fast Propagation of Firing Rates through Lay-
ered Networks of Noisy Neurons,” J. Neurosci., vol. 22, no. 5, pp. 1956–1966, Mar. 2002.

UnERD
2O19 • 56
AUG 13
D: Materials Science
D1: Towards the replacement of conventional platinum cathode for low-temperature fuel cells: A
novel approach for the development of non-precious metal cathode
Shreya Gupta1,2, Jayesh Manohar Sonawane1,2, Yu-Ling Cheng1,
1
Centre for Global Engineering, University of Toronto; 2Department of Applied Sciences and Engineering, University of Toronto

The catalyst plays a significant role in low-temperature fuel cells, such as polymer electrolyte membrane fuel cells
(PEMFC) and microbial fuel cells (MFC). Platinum has been extensively used for PEMFCs and MFCs due to its
high oxygen reduction potentials (ORP), stability, and good electrokinetics. In recent years, numerous efforts have
been made to replace the precious platinum catalyst with its non-precious counterpart due to economic viability. We
studied the use of novel polyaniline with non-precious nickel based metal composites as a potential low-cost
replacement for the conventional platinum cathode. The incorporation of non-precious metals on top of the polyani-
line coating enhances the catalytic activity for the oxygen reduction reaction and thus increases the production of the
end- product (electricity). The precise quantity and composition used yields around 150% higher exchange current
density when compared with the conventional platinum coated cathode made up of a 0.5 mg cm-2 platinum loading.
An electron microscope investigated the morphology of the polyaniline-based coating in order to gain a better
understanding of the performance of the cathodes. It was observed that the highly conducting nano-fibrous structure
(40-50 nm) for the fabricated composite attributes an enhanced charge transfer from an external circuit to the reac-
tion center. This study demonstrates that the proposed polyaniline based non-precious cathode can be adopted as a
suitable low-cost replacement for the platinum cathode. This electrode can be used in electrochemical devices such
as MFCs, hydrogen fuels cells and batteries.

D2: Development of Carbon Fibre Composite Tetrahedral-Octahedral Truss Structures with Effi-
cient Mechanical Behaviour
Shengyu (Brad) Sun, Calvin Moes, Glenn Hibbard
1
Department of Materials Science and Engineering, University of Toronto

Tetrahedral-octahedral truss structures consist of two tetrahedra with several octahedra in between. This structure
allows material to shift away from the neutral axis, thus making it highly efficient in certain load conditions com-
pared to conventional designs. Carbon fibre (CF) composites have high stiffness and specific strength, which
increases the efficiency in such structures. The aim of this study is to construct ultra-lightweight CF trusses with
such structure that can withstand a relatively high compression load. In the fabrication process, a single CF strand is
saturated with epoxy resin before being laid around a 3D printed wax pre-form with designated channels. The wax
is melted out after the composite structure solidifies, producing a hollow truss structure. Since the structure is likely
to fail by Euler buckling, the compression test results are the focus. To investigate the compression responses of this
structure, three sets of trusses and single struts, consisting of 2, 4, and 6 layers of CF strands were prepared. The
samples were tested under the Euler buckling modes of pinned and fixed ends. Test results show that the critical
buckling load of a 2.7-gram 6-layer CF truss was 863 N, on average. Future plans include constructing hierarchical
truss structures that can further exploit the mechanical efficiency. It is important to optimize the structure by balanc-
ing the failure modes (i.e. strut buckling, lower-level truss buckling, and overall buckling of the structure). The
results from this study are expected to assist in predicting the failure modes of the hierarchical structure.

57 • UnERD
2O19
AUG 13
D3: Ethylene Glycol as Additive in Na2SO4-PAM Electrolyte for Low Temperature Supercapaci-
tors
Gary Liu (MSE 2T1)1, Alvin Virya1, and Keryn Lian1
1
Department of Materials Science and Engineering, University of Toronto

Supercapacitor is an energy storage device that provides high energy and power densities with longer service life.
Polymer electrolytes, with ion conductors held in polymer host matrices, are developed to replace traditional liquid
electrolytes. They have advantage in safety, flexibility and light weight, properties desirable for wearable electronic
devices [1]. Sodium sulfate (Na2SO4)-polyacrylamide (PAM) system has been demonstrated to be a promising
polymer electrolyte, displaying wide voltage window, high ionic conductivity, excellent cycling capability and low
cost [2]. However, its conductivity and resulting capacitance drastically decrease below 0°C, limiting its application
in low temperature environment, such as in the subarctic climate of Canada.

This research aims to improve the low temperature performance of Na2SO4-PAM system through the addition of
ethylene glycol as additive. Electrochemical performance of electrolytes containing different concentrations of
ethylene glycol was characterized using cyclic voltammetry and AC impedance and tracked at ambient conditions
over time. Their low temperature applications were emulated in an environmental chamber between -30 and 0°C.
Material properties of these polymer electrolytes were investigated using techniques such as x-ray powder diffrac-
tion (XRD) and Fourier-transform infrared spectroscopy (FTIR). The cell with 10000:1 ethylene glycol to PAM
molar ratio was able to maintain capacitive property down to -20 degree Celsius, which is much better than their
binary baseline. This signified the effect of introducing ethylene glycol. The improvement in capacitance is also
coupled with a decrease in charge-transfer resistance, another positive feature of this additive. One future direction
may be to investigate the mechanism behind this phenomenon.

[1] H. Gao and K. Lian, "Proton-conducting polymer electrolytes and their applications in solid supercapacitors: a
review," Royal Society of Chemistry, vol. 4, pp. 33091-33113, 2014.

[2] A. Virya, J. Abellaa, A. Grindala and K. Lian, "Na2SO4-Polyacrylamide Neutral pH Polymer Electrolytes for
Electrochemical Capacitors," in The Electrochemical Society, Dallas, 2019.

UnERD
2O19 • 58
AUG 13
D4: Highly-conductive and Ultra-elastic Polyether Block Amide/Carbon Nanotube
Nanocomposites
Qiwei Zhao, Dr. Jun Wang
1
Department of Mechanical and Industrial Engineering, University of Toronto

Stretchable and conductive polymer nanocomposites have shown great potential for interactive electromechanical
applications, such as sensors, actuators, and electronics [1]. The materials made of polymers are light-weight and
easy to process. The polymer nanocomposites are composed of two phases — the continuous matrix, and the
dispersed filler, which is surrounded by the matrix. However, the conductivity of the polymer nanocomposite is
generally much lower than that of the conductive filler since most of the polymer matrices are electrical insulative,
resulting in increased contact resistance between the fillers. This study aims to test the composite with carbon nano-
tubes(CNTs) that are highly conductive material as the conductive fillers, and the ionically conductive polyether
block amide (PEBA) elastomer, which has relatively low conductivity, is used as the polymer matrix. It is hypothe-
sized that by using a conductive matrix, the contact resistance between CNTs can be significantly reduced. There-
fore, it is expected to generate highly conductive PEBA/CNT nanocomposites with excellent elasticity for wearable
electronics and electromagnetic interference shielding applications. Currently, the preliminary results show that the
nanocomposite conductivity can be increased by 10-100 times when the CNTs are implemented, as compared to
pure PEBAs. With ~ 10vol% CNT in the composite, the conductivity can reach 30S/m. The next steps of the project
are to use different types of PEBA, to add a nucleating agent to modify the PEBA crystal structures, and to imple-
ment different manufacturing processes to increase the conductivity further.

[1] R.Paradiso,” Smart Fabrics and Interactive Textile: State of the Art and Future Challenges” IEEE, Mar 30,
2009. [Online]. Available: ]https://ieeexplore-ieee-org.myaccess.library.utoronto.-
ca/courses/content/EW1102/EDP102Notes.pdf [Accessed July,26,2019]

59 • UnERD
2O19
AUG 13
D5: Utilization of Creep Test to Characterize the Mechanical Properties of a Sn-Ag-Cu-Bi Pb-Free
Solder Alloy
Juan Pablo Tupac Yupanqui Cardoso1,2, André Delhaise1,3, Fae Azhari3,4, Doug D. Perovic1

Materials Science & Engineering, University of Toronto, 2Decisionics Lab, University of Toronto; 3Celestica, Inc.; 4Civil and Mineral
1

Engineering, University of Toronto

In 2006, the Restriction of Hazardous Substances (RoHS) restricted the use of lead (Pb) in electrical and electronic
equipment, causing a crucial need to develop a Pb-free solder solution within the electronics industry [1]. Currently,
SAC305 (Sn-3.0Ag-0.5Cu) solder is used on printed circuit boards and other board materials that can withstand high
process temperatures. However, many electronics industries such as aerospace, military, and automotive have found
that SAC305 is not idea [2]. Some of these problems include the high cost of silver, poor mechanical properties such
as poor drop performance and high process temperature, susceptibility to pad cratering, and increased risk of system
failures from tin whiskers [2]. This led to the study in the reliability of adding bismuth (Bi) to the alloy as it reduces
the melting temperature, improves thermomechanical properties, and the possibility of reducing the tendency of
whisker formation [2].

The current research will focus on investigating the mechanical properties of the “Violet” alloy (Sn-2.25Ag-0.5-
Cu-6.0Bi), one of the seven proposed alloys from past research with Celestica and UofT [3]. Creep tests will be
conducted on the alloy, where the collected data will be used for the construction of a thermomechanical fatigue
model. This research will follow ASTM E9 methodology where some samples will be aged beforehand (125°C for
24 hours), subjected to one of several temperatures (-25°C, 25°C, 75°C, 125°C), and different loads (5MPa, 10MPa,
15MPa, 25MPa, 35MPa, 40MPa, 45MPa, 50MPa).

[1] "Lead-Free and RoHS Summary - Maxim", Maximintegrated.com, 2019. [Online]. Available: https://www.max-
imintegrated.com/en/support/emmi/lead-free-rohs-summary.html. [Accessed: 27- Jul- 2019]

[2] P. Snugovsky, S. Bagheri, M. Romansky, D. Perovic, L. Snugovsky and J. Rutter, "New Generation Of Pb-Free
Solder Alloys: Possible Solution To Solve Current Issues With Mainstream Pb-Free Soldering - Journal of SMT",
Smta.org, 2012. [Online]. Available: https://www.smta.org/knowledge/journal_detail.cfm?ARTICLE_ID=206.
[Accessed: 27- Jul- 2019]

[3] A. Delhaise, P. Snugovsky, I. Matijevic, J. Kennedy, M. Romansky, D. Hillman, S. Meschter, J. Juarez, M.

Kammer, I. Straznicky, L. Snugovsky, D. Perovic and D. Adams, THERMAL PRECONDITIONING, MICRO-
STRUCTURE RESTORATION AND PROPERTY IMPROVEMENT IN Bi-CONTAINING SOLDER ALLOYS.

UnERD
2O19 • 60
AUG 13
E: Electrical and Computer
E1: Investigation of Massive MIMO Channel Correlations Across Frequencies
Russell Brown1, Raviraj Adve1
1
Edward S. Rogers Sr. Department of Electrical and Computer Engineering, University of Toronto;

One of the major technologies intended to realize the 5th generation of cellular networks (5G) is Massive Multi-
ple-Input and Multiple-Output (MIMO) antenna arrays [1]. Massive MIMO has been well studied in Time Division
Duplex (TDD) systems due to the assumption of channel reciprocity, the idea that the downlink channel should
resemble the uplink channel, an assumption that is no longer valid in Frequency Division Duplex (FDD) systems
[2]. Traditional MIMO FDD systems gain channel state information through feedback and training, but the
increased scale of Massive MIMO makes this impractical. This is because the resources required for downlink train-
ing scale with the large number of transmit antennas [3]. As many existing cellular systems are FDD [2], a Massive
MIMO FDD system would be one step towards enabling 5G technology in those cellular systems. In order to under-
stand the characteristics of Massive MIMO FDD, simulation software meeting standards specified by 3GPP (3rd
Generation Partnership Project, a wireless standard creation organization) was used. The software was used to simu-
late a Massive MIMO uplink at a base frequency, and downlink channels across several higher frequencies, for a
specific antenna arrangement. These channels were then used to calculate averaged correlation matrices between the
uplink and downlink across frequencies. Expected results are a channel that becomes less correlated as the frequen-
cy difference between downlink and uplink increases, though the simulation results are not yet available. Once the
correlation is better understood, the information will be used to determine feasible setups for Massive MIMO FDD
systems.

[1] Ericsson, “Advanced antenna systems for 5G networks,” Ericsson, 2018. [Online]. Available:https://ww-
w.ericsson.com/en/white-papers/advanced-antenna-systems-for-5g-networks. [Accessed: July. 27, 2019]

[2] M. B. Khalilsarai, S. Haghighatshoar, G. Caire, “How to Achieve Massive MIMO Gains in FDD Systems?”,
Proceedings of IEEE International Conference on Communications, Jun. 2018.

[3] D. Wang, X. Y. Wang, X. Yang, Y. Zhao, W. Wan, "Design of Downlink Training Sequences for FDD Mas-
sive MIMO Systems", Proceedings of IEEE International Conference on Communications, Jun. 2015.

61 • UnERD
2O19
AUG 13
E2: High Precision Optical Delay Measurement via Hong-Ou-Mandel Interference of Adjustable
Frequency Comb Filtered Spectral Biphotons
Takahiro Tow1, Changjia Chen1, Arash Riazi1, Li Qian1
1
Department of Electrical and Computer Engineering, University of Toronto

Entangled Photon states produced in optical fiber are an emerging resource for metrology and communications as
they offer better performance than their classical counterparts. Hong-Ou-Mandel interference (HOM) is a two-pho-
ton quantum interference phenomenon, used to measure optical delays (or temporal distinguishability), in which the
interference of the photons is dependent on their distinguishability: the coincidence probability is zero for identical
photons. Greater precision in delay measurement using HOM interference is typically achieved by increasing the
input biphoton bandwidth. Alternatively, we investigate a novel method which uses spectrally-entangled biphotons
with an adjustable discrete comb filter frequency profile. Greater precision is achieved from the resultant temporal
quantum beating phenomenon, limited theoretically by a higher-resolution Quantum Cramer-Rao bound [1]. Broad-
band entangled photons ( 90nm bandwidth) are produced in periodically-poled Silica fiber via Type-II spontaneous
parametric down conversion [2]. An optical delay is induced by an adjustable motorized delay line in one arm of a
Mach-Zehnder interferometer. A pulse shaper is inserted into the other arm of the interferometer for post-selection
of discrete frequency spectra, producing the beating phenomena in the coincidence counts detected using single
photon counters. The relationship between difference frequency of the shaped frequency combs and its resultant
beating pattern will be investigated. The ultra-precise measurement of optical path delay is a necessary advancement
towards the separation of hyper-entangled states (entanglement in >1 degree of freedom) in fiber. Hyperentangle-
ment and pulse-shaping are promising quantum communication resources as they each add another degree of free-
dom for data encoding.

[1] Y. Chen, M. Fink, F. Steinlechner, J. P. Torres, and R. Ursin, “Hong-Ou-Mandel interferometry on a biphoton
beat note,” npj Quantum Information, vol. 5, no. 1, 2019.

[2] C. Chen, E. Y. Zhu, A. Riazi, A. V. Gladyshev, C. Corbari, M. Ibsen, P. G. Kazansky, and L. Qian, “Compen-
sation-free broadband entangled photon pair sources,” Optics Express, vol. 25, no. 19, p. 22667, 2017.

UnERD
2O19 • 62
AUG 13
E3: Organ Classification of Histological Whole Slide Images by Aggregating Patch Level Visual
Features
Danial Hasan1, Mahdi Hosseini2, Konstantinos Plataniotis2
1
Division of Engineering Science, University of Toronto; 2Department of Electrical and Computer Engineering, University of Toronto

Histology is the study of tissue structure at the microscopic level. Traditionally, histology is conducted using micro-
scopes to observe glass slides of tissue samples. However, developments in Whole Slide Imaging (WSI) have
enabled digital scanning of glass slides for efficient storage and greater usability. The field of computational pathol-
ogy aims to develop tools that can analyze digital WSIs to increase the accuracy and productivity of pathologists in
diagnosing disease within histological samples. Until now, most research in computational pathology has been
constrained to specific diseases or organs due to the lack of generalized histological datasets.

This work aims to develop a generalized encoding to classify the originating organ of histology WSIs, by aggregat-
ing visual features observed on patches of WSIs. The performance of encodings is measured by the accuracy of
organ classification of origin for histological WSIs using a nearest neighbour classification. We demonstrate that
visual features extracted by Convolutional Neural Nets can be aggregated using Content Based Image Retrieval
methods such as Bag-of-Words[] and VLAD [] to yield encodings that achieve good performance for organ classifi-
cation on WSIs that have mostly healthy tissue. Of note, our custom CNN HistoNet, trained to classify histological
tissue types in general histology patches, outperforms state-of-the-art CNNs trained on both ImageNet and histology
data with an encoding of much smaller size. Development of such an encoding is highly applicable in CBIR systems
for pathologists in finding similar WSI regions and for integration into novel Laboratory Information Systems (LIS).

[1] G. Csurka, C. R. Dance, L. Fan, J. Willamowski, and C. Bray, “Visual cat-egorization with bags of
keypoints,” inIn Workshop on Statistical Learningin Computer Vision, ECCV, 2004, pp. 1–22.

[2] H. J egou, M. Douze, C. Schmid, and P. P erez, “Aggregating local descrip-tors into a compact image repre-
sentation,” in2010 IEEE Computer Soci-ety Conference on Computer Vision and Pattern Recognition, Jun. 2010,pp.
3304–3311.doi:10.1109/CVPR.2010.5540039.

63 • UnERD
2O19
AUG 13
E4: Unsupervised EEG Signal Segmentation with Exponentially Decaying Memory and Deep
Autoencoders for Label Generation
Shujian Qian1,2, Gerard O’Leary1,2, Dr. Taufik A. Valiante1,3,4
1
Krembil Research Institute, University Health Network; 2Dept. of Electrical and Computer Engineering, University of Toronto; 3Institute
of Biomaterials and Biomedical Engineering, University of Toronto; 4Division of Neurosurgery, Dept. of Surgery, University of Toronto

Epilepsy is a chronic disorder of the nervous system characterized by the recurrent onset of brain states known as
seizures. These states involve the abnormal synchronized firing of a large population of neurons and can cause
patients to lose control of voluntary function. Closed-Loop electrical neurostimulation is a promising approach for
seizure reduction that is being actively investigated. This process requires a seizure state to be detected through the
analysis of electroencephalographic (EEG) data from electrodes implanted in the brain. Recent advances in machine
learning have allowed for the use of pre-recorded EEG signals to train a model to detect the imminent onset of a
seizure. Accurate models require accurate segmentation labels, but seizure data labels contain a high level of human
error. In this research, we propose an unsupervised learning approach to overcoming the limitations of inaccurate
labels. Firstly, a layer of exponentially decaying memory (EDM) units extracts the temporal features from the EEG
signal [1]. Afterward, a deep autoencoder encodes the extracted features into multi-dimensional latent vectors which
are then clustered to produce accurate labels. Early results have shown that the algorithm can generate clean clusters
of brain states. Next steps include using these clusters to train computationally efficient models for real-time seizure
prediction and early interruption with stimulation. Furthermore, the clusters can allow the detection of seizures and
enhance our understanding of how brain activity evolves in the lead-up to a seizure.

[1] G. O'Leary, D. M. Groppe, T. A. Valiante, N. Verma, R. Genov, "NURIP: Neural interface processor for
brain-state classification and programmable-waveform neurostimulation", IEEE J. Solid-State Circuits, vol. 53, no.
11, pp. 3150-3162, Nov. 2018.

UnERD
2O19 • 64
AUG 13
F: Mechanical and Industrial
F1: Enhancing the sound absorption property of polypropylene microcellular foams with
graphene for automobile industry
Jianfei Pan, Jiayi Liu, Yun-Seok Jun, and Chul B. Park

Microcellular Plastic Manufacturing Laboratory, Department of Mechanical and Industrial Engineering, University of Toronto, 5 King’s
College Road, Toronto, Ontario, Canada M5S 3G8

Automobile usage greatly contributes to the noise endured by humans every day. Lightweight polymeric foams with
effective acoustic properties are designed to reduce noise from coming into the vehicle’s cabin for passengers.
Industrially, there has been extensive research and application of the sound absorbing graphene enhanced polyure-
thane (PU) foams. However, currently there is a gap in understanding the effect of graphene on the sound absorption
coefficient of polypropylene (PP) foams. Compared to polyurethane foams, polypropylene foams are advantageous-
ly recyclable and economic. [1] It’s hypothesized that the addition of graphene would drastically change the cell
morphology of PP foams in terms of cell sizes and densities so as to enhance their acoustic properties. Therefore, an
effective and environmentally friendly sound absorbing foam can be produced. In this project, PP foams with differ-
ent graphene contents are produced via foam injection molding technology. The sound absorption coefficients are
obtained from the ASTM E1050 Transfer-Function method using an impedance tube. [2] The current results show
that graphene content enhances the absorption coefficients. Future missions will focus on the open cell contents and
scanning electron microscope (SEM) examinations to microscopically analyze the impact of graphene on cell mor-
phology. This study will demonstrate the effect of adding graphene on enhancing the acoustic properties of PP
foams, while investigating the mechanism behind this phenomenon in terms of cell morphology changes and open
cell contents, for understanding and developing future acoustical materials with outstanding sound absorption capa-
bility.

[1] Tspace.library.utoronto.ca. (2019). [online] Available at: https://tspace.library.utoronto.ca/bit-

stream/1807/69246/3/Jahani_Davoud_201506_PhD_thesis.pdf [Accessed 26 Jul. 2019].

[2] Compass.astm.org. (2019). ASTM International - Compass Login. [online] Available at: https://com-
pass.astm.org/EDIT/html_annot.cgi?E1050+12#s00040 [Accessed 26 Jul. 2019].

65 • UnERD
2O19
AUG 13
F2: Reliability Model for a transit bus Intelligent Transportation System
Sohaira Khan1, Professor Viliam Makis1
1
Department of Mechanical and Industrial Engineering, University of Toronto

Public transit is a cost-intensive operation where a significant portion of the total cost is attributed to its mainte-
nance. An optimal maintenance policy that aims to reduce these costs requires reliability models. This study
focuses on the development of a reliability model for an Intelligent Transportation System (ITS) found in public
transit buses. Reliability modelling of multi-component systems is an emerging field of research and a reliability
model for an ITS has not been developed in the literature. The ITS is a combination of sub-systems. Pareto Charts
have been used to identify The Surface Vehicle Safety Camera System and Surface Automated Stop Announce-
ment System as sub-systems with the highest number of failures. To develop the reliability model, the impact of
imperfect repair actions on the rate of occurrence of failure (ROCOF) is studied for each individual component
within the identified sub-systems. Component failures are modelled using Non-Homogenous Poisson Processes
(NHPP) where the ROCOF is the intensity of NHPP and it varies with time. Maximum Likelihood Estimation
(MLE) using real failure data supplied by a public transit operator is used to determine the estimates of parameters
that define the ROCOF. MLE equations have been developed and different intensity functions for the reliability
model have been investigated. The obtained parameters estimates are then used to build the reliability model. The
model developed will contribute to the understanding of reliability of multi-component systems. Future work
includes the development of an optimal maintenance policy using the reliability model that reduces public transit
operation costs.

F3: The Application of a Combined Solar Chimney and Wind Catcher System for the Natural
Ventilation of High-Rise Multi-Unit Residential Buildings
Yanxi Li1, Jamie P. Fine2, Marianne F. Touchie3

Mechanical & Industrial Engineering, University of Toronto; 2Mechanical and Industrial Engineering, University of Toronto; 3Jointly
1

Appointed in Civil & Mineral Engineering and Mechanical & Industrial Engineering, University of Toronto

Natural ventilation techniques are being considered and studied as a way to maintain a healthy and comfortable
indoor environment while reducing energy consumption [1]. Two natural ventilation techniques, solar chimneys
and wind catchers, are being investigated as a combined system, which both take advantage of renewable energy
to generate airflow. There is a lack of literature on the study of combined wind catcher and solar chimney systems,
especially in the context of high-rise multi-unit residential buildings (MURBs). This lack of study therefore
hinders the future adoption of this natural ventilation strategy, and given the renewable nature of this system,
serves to motivate this research. In this paper, a combined solar chimney and wind catcher system is analyzed for
use in a multi-story building, and design iteration is carried out to optimize system performance. A custom numeri-
cal method was developed in MATLAB to characterize the system using the balance of pressure losses and energy
gains, and to solve the parallel flow network. This model provides insight into how the system performs different-
ly in a MURB due to the increased stack height of the solar chimney and longer distance between air inlets and
outlets. An experimental trial using a building model will be conducted to validate the results of this analysis.
Using these results, the benefits of the proposed system are determined through a comparison against a conven-
tional mechanical ventilation. In hot and temperate climates, this system has the potential to reduce building
energy consumption [1], resulting in lower emissions and a decreased carbon footprint.

[1] P. A. O. Ivan Oropeza-Perez, "Active and passive cooling methods for dwellings: A review," Renewable and
Sustainable Energy Reviews , vol. 82, pp. 531-544, 2018.

UnERD
2O19 • 66
AUG 13
F4: Experimental Study and Simulation of the Thermally Conductive Dielectric Polymeric Com-
posites
Shang Shi, Azadeh Zandieh, and Chul B. Parks
1
Microcellular Plastics Manufacturing Laboratory (MPML), University of Toronto

Thermally conductive yet electrically insulating materials serve an important role in microelectronic packaging
and thermal management applications. The current trend towards miniature electronics require advanced materials
with a high thermal conductivity and electrical resistivity. Currently, polymeric composites are at the center of
attention for the purposes mentioned above due to its relative low costs, ease of processability and desirable
properties. Our study aims to develop such a polymeric composite with polycarbonate (PC) being the matrix and
hexagonal boron nitride (h-BN) as the filler. Based on their individual properties, we expect that a combination of
these two materials will create a composite with a high enough thermal conductivity and electrical resistivity
useful in microelectronics. We sought the optimal processing conditions that create a continuous filler network
before finding the optimal filler weight percentage that suits our needs. First, electron and optical microscopies
were used to characterize the morphology of the composites. Then, we studied how the filler morphology can
affect the composite’s thermal conductivity by using spherical and platelet shaped h-BN agglomerates. From all
the experimentation, we confirmed the increase in thermal conductivity caused by the conductive path formed by
the h-BN but have yet to find any correlation between the morphology of fillers and the thermal conductivity. We
verified the measured thermal conductivities with a numerical simulation through COMSOL. Currently, we have
moved on to studying how BNNT as the filler affects the composite’s thermal conductivity and dielectric proper-
ties using similar methodology as discussed above.

F5: Nestle’s Underpayment Claim Classification via Logistic Regression

Ruiyang Li, Jiaru Li, Aloagbaye Momodu, Professor Chi-Guhn Lee
1
Department of Mechanical & Industrial Engineering, University of Toronto;

Nestle is one of the world’s largest food and beverage companies. It generates billions in revenue. However,
deduction claims cause it to lose about 25 million per year.
The deduction claims are claims sent to Nestle when late delivery, insufficient goods or unsatisfactory service
level occurs. When the above situations happen, Nestle will prioritize the order of investigation. At the moment,
Nestle is processing underpayment claims manually, so it is low-efficient and time-consuming. The objective of
this research is using machine learning techniques to create an automated process which efficiently optimizes the
processing procedure.

The first step to generate an ideal result is data cleaning and exploratory data analysis (EDA). Logistic regression
is proposed to analyze past years’ data. From the Definition of Logistic regression, Success rate = 1/(1+e-t), where
t is a combination of weighted features selected during the EDA stage. At last, the probability of disapproving the
claim is generated. After building the rudimentary model, K-Fold is used to separate testing data and training data.
F1-score and ROC-AUC score are used to measure the accuracy of the model. The time the model used will be
measured and compared with manual processing time to show its efficiency. The model is expected to produce
similar results as this year’s data. Accuracy of the model will be measured by the similarity of output result and
actual result. Since Logistic regression generates continuous results, the result might not be ideal for Nestle.
Clustering and decision tree are considered as alternative models to produce the ideal result for Nestle. Via apply-
ing the model, Nestle could identify claims more efficiently, save both labor and expense cost. It could potentially
recover millions of revenue thought the project.

67 • UnERD
2O19
AUG 13
F6: A modified deep regression neural network for American option pricing
Mingkun Wang1,Aloagbaye Momodu1,2, Chi-Guhn Lee1,2
1
Department of Mechanical and Industrial Engineering, University of Toronto; 2Dynamic Optimization & Operations Management
Laboratory, University of Toronto;

American options are financial instruments which guarantee the right for the option holders to buy or sell an
underlying asset at a predetermined price in the future. It provides a safer form of investment compared to Europe-
an options since holders can exercise at any time before maturity. However, the nonlinearity and stochasticity in
selecting an optimal stopping time make American options difficult to be priced. Thus, it creates the interest of
developing a quick, efficient, and accurate option pricing method which can be applied to general financial data.
In this project, the team aims to develop a multi-layered regression neural network model, which predicts option
prices with minimal error and cost. Considering how traditional European options can be priced with the
Black-Scholes PDE, a sequential model is built with three hidden layers and the same parameters of the
Black-Scholes equation[2]. To fit in our regression model, datasets are selected, normalized and split up into
training and validation sets[3]. With the new algorithm, the model is trained with implied volatility data to develop
standard parameter weights and predicts results from realized volatility data. Two statistics, root mean squared
error (RMSE) and R-Squared, are introduced to better represent the result accuracy. Based on the metrics, the
model has an RMSE value of 2.23 and an R-Squared value of 0.966, which shows great accuracy of the error-free
model with given datasets. In the future, the current model will be adjusted accordingly for generalized data and
more accurate results.

[1] R. Zvan, P. Forsyth, and K. Vetzal, “Penalty methods for American options with stochastic volatility,”
Journal of Computational and Applied Mathematics, vol. 91, no. 2, pp. 199–218, 1998.

[2] F. Black and M. Scholes, “The Pricing of Options and Corporate Liabilities,” Journal of Political Economy,
vol. 81, no. 3, pp. 637–654, 1973.

[3] U. Anders, O. Korn, and C. Schmitt, “Improving the pricing of options: a neural network approach,”
Journal of Forecasting, vol. 17, no. 5-6, pp. 369–388, 1998.

UnERD
2O19 • 68
AUG 13
F7: Traffic Display and Drivers’ Experience Influencing Drivers’ Distraction Behaviours in Auto-
mated Vehicles
XueGe Huang, Xiaoou Li, Dengbo He, Dina Kanaan, Birsen Donmez
1
Department of Mechanical and Industrial Engineering, University of Toronto, Human Factor and Statistics Lab (HFASt)

Driver distraction is one of the major threats to driving safety in non-automated vehicles. The same applies to
automated vehicles where drivers are more likely to engage in non-driving-related tasks (NDRT), which will delay
drivers’ responses in critical situations. In order to support the transfer from manual to automated driving, a vari-
ety of supporting displays (e.g., take-over request or automation capability display) were proposed in automated
vehicles. However, to our knowledge, no studies have investigated how these displays may influence NDRT
engagement in automated vehicles. In this study, we compared NDRT engagement behaviors of drivers when
provided different levels of displayed information in automated vehicles. Two types of displays were compared,
Display A with automation capability information and take-over request in case of emergency, and Display B,
which provides surrounding traffic information in addition to the information in Display A. Considering how
experienced drivers were found to show relatively conservative NDRT engagement behaviors in automated vehi-
cles, 24 novice and 24 experienced drivers were recruited and a between-subject experiment design was used[1].
Our results show that both novice and experienced drivers demonstrated a relatively higher level of NDRT
engagement with Display A comparing to both baseline (without display) and Display B. These results indicate
that drivers tend to trust the automation system the most with the help of only automation capability and the
take-over request. This suggests that, future interface design in automated vehicles should consider the side-effect
of different information on drivers’ NDRT engagement.

[1] D. He and B. Donmez, “Influence of Driving Experience on Distraction Engagement in Automated Vehicles”,
Transportation Research Record: Journal of the Transportation Research Board, p. 036119811984347, 2019.

F8: MoS2 Lubricant Performance in Low Earth Orbit Temperatures

Khaled Gaber1, Peter Serles1, Simo Pajovic1, Tobin Filleter1
1
NanoMechanics and Materials Laboratory, Department of Mechanical and Industrial Engineering, University of Toronto

As space exploration evolves further, the duration and scope of space missions has also evolved to more complex
systems and to the magnitude of multi-year missions. One major difficulty associated with increased space travel
durations is the deterioration of mechanical parts due to the degradation of the lubrication used. Prior research has
been conducted towards the development of new coatings for space related applications, with a material of interest
being molybdenum disulfide (MoS2). The properties of MoS2 include ultra-low friction coefficient values and
good performance and wear life in vacuum, making it ideal for use within space travel. Current literature has
investigated the role of increased temperature, however few have done so with levels of sensitivity possible with
Atomic Force Microscopy (AFM), and there exists no literature utilizing the AFM in studies with thermal cycling.
The effect of temperature on MoS2 will be investigated using AFM and a custom high-temperature environmental
control box. MoS2 will be exposed to temperatures mimicking the high-temperature environment experienced in
Low Earth Orbit by mechanisms such as the International Space Station. Tribological tests will be done in-situ
from 25 to 120°C and ex-situ temperature cycling from -120 to 120°C in air and humid nitrogen conditions.
Temperature dependence above a transition temperature is predicted to correlate with desorption of water. The
transition temperature is expected to be dependent on the presence of humidity during experimentation. Environ-
mental humidity conditions and cycle repetitions are expected to be significant influences on the size of the tribo-
logical hysteresis loop.

69 • 2O19 UnERD
AUG 13
F9: Analysis on Traffic Display Influencing Drivers’ Distraction Behaviors in Automated Driving
Xiaoou Li, Xuege Huang, Dengbo He, Dina Kanaan, Birsen Donmez

Department of Mechanical and Industrial Engineering, University of Toronto, Human Factor and Statistics Lab (HFASt)

The prevalence of automation allows more non-driving-related tasks (NDRT) engagement, which reduced antici-
patory actions for both experienced and novice drivers and potentially impair take-over performance in case of
emergency [1]. Previous research has proposed a variety of displays that may support transfer of control in auto-
mated vehicles [2]. However, to the best of our knowledge, no research investigated how different sources of
information may influence drivers’ NDRT engagement. In this research, a driving simulator study was conducted
with 24 experienced drivers and 24 novice drivers [3]. Each participant completed 4 drives in an SAE-level 2
automated vehicle and was allowed a visual-manual NDRT. A between-subjects design was used so that each
participant experience one of the three types of traffic display in the experiment, i.e., no display, Display A with
automation capability information and take over request, and Display B with the information of surrounding traffic
information in addition to the information provided in Display A. NDRT engagement is measured by gaze posi-
tion, rate of manual interaction, rate of long off-road glances (>2 s), average duration of each glance and percent-
age of time spent on the secondary task. The preliminary results show that both experienced and novice drivers
with Display A showed a relatively higher level of NDRT engagement, indicating the presence of automation
capability information and the take-over request only may lead to over reliance on automation. This research
indicates that future in-vehicle display design should consider their side effect on drivers’ behaviors.

[1] D. He and B. Donmez, (2019) “The Effects of Distraction on Anticipatory Driving”. Proceedings of the
Human Factors and Ergonomics Society Annual Meeting, 62(1), 1960–1964.

[2] P. Stahl, B. Donmez, and G. A. Jamieson, “Supporting anticipation in driving through attentional and interpre-
tational in-vehicle displays,” Accident Analysis & Prevention, vol. 91, pp. 103–113, 2016.

[3] D. He and B. Donmez, “Influence of Driving Experience on Distraction Engagement in Automated Vehicles”,
Transportation Research Record: Journal of the Transportation Research Board, p. 036119811984347, 2019.

UnERD
2O19 • 70
AUG 13
F10: Effect of Polymer Molecular Structure on GnP and Graphite Exfoliation
Kelly Chu, Dr. Jun Wang, Prof. Chul B. Park

Department of Mechanical and Industrial Engineering, University of Toronto

Graphene is a monolayer of carbon atoms arranged in a hexagonal structure with remarkable electrical, thermal,
and mechanical properties. However, the commercial graphene materials, such as graphene nano-platelets (GnPs),
exhibit inferior properties because they are composed of multiple graphene layers.

Graphene/polymer composites is a way to utilize the extensive properties of graphene in practical use. Therefore, a
greater knowledge of how the polymer molecular structure in the composite influences graphene exfoliation is
crucial in determining a cheaper and more efficient way to produce composites with high-quality graphene. The
purpose of the experiment is to find a correlation between polymer molecular structure and exfoliation, which will
enable more advanced manufacturing strategies in producing graphene composites with extensive applications

The experimental procedure involves preparing a wide variety of polymer and graphene composites. Polymers are
divided into categories according to molecular architecture (linear or branched) and molecular weight/viscosity.
Different forms of GnP and commercial grade graphite powder are used as graphene precursors. To determine the
degree of exfoliation, the composite will undergo exfoliation treatment using a supercritical fluid assisted foaming
process, then will be scanned and analyzed through the X-Ray Diffraction (XRD) analysis. A lower diffraction
intensity of the peak for graphene layer-layer stacking indicates a more effective exfoliation.

Preliminary results show that a linear polymer is more effective in facilitating GnP or graphite exfoliation as
compared to the branched ones. Future steps include varying the exfoliation temperature and pressure to deter-
mine how they affect the graphite exfoliation in the composites.

F11: Effect of CAD collaboration methods on the quality of CAD models

Hamza Arshad1, Vrushank Phadnis2, Alison Olechowski1
1
Mechanical and Industrial Engineering, University of Toronto; 2Massachusetts Institute of Technology

Computer-Aided Design (CAD) processes have historically been carried out by design teams by having one static
file copied between different workstations to allow each individual member to propose design changes. New
collaborative CAD software, however, allows designers to collaborate in real time and propose changes on one
dynamic file. Research is needed to understand the optimal methods to organize CAD collaborators to produce
high quality CAD models using this new collaborative CAD software. This paper explores two different methods
of CAD collaboration to understand how quality is affected. Pairs using either of the two methods are given a
CAD file to implement up to fourteen changes in forty-five minutes
.
The first method includes pairs using a shared screen to collaborate on a file with one person driving the software
at a time, while the second method has pairs collaborating on the same file with the opportunity to parallelize on
tasks. Quality is divided into two different categories: to what degree did users follow the feature requirements for
each change and adherence to standard CAD practices (ie: constraining sketches). Both categories are measured
using a common scoring matrix between participants. It is expected that the first method would produce higher
quality CAD models as it allows more direct collaboration between pairs who work on one change at a time (no
parallelization) due to the shared screen.

71 • UnERD
2O19
AUG 13
 G: Civil/Mineral
G1: Improvement of Rock Fragmentation Measurements utilizing Aerial Drone-captured Image
Recognition and Deep Neural Networks
Tianyu(Amily) Xu, Thomas Bamford, Dr. Kamran Esmaeili
1
Lassonde Institute of Mining, University of Toronto

Rock fragmentation data is crucial to the prediction and monitoring of blasting activities in mining and quarry opera-
tions. The measurement of fragmentation provides information on the economically significant size range of a defin-
able volume of broken rock, contains oversize, mid-range, and fine grains [1]. The total production rate and product
value of the overall processing system is significantly dependent on rock particle size distribution.

Some conventional computer vision techniques, including threshold and watershed algorithms, have been previous-
ly developed for the automated measurement of rock segmentation. But the accuracy of applying these techniques
are restricted to a number of external conditions, some of which maybe common in outdoor environments but detri-
mental to the precise interpretation of raw data, such as ambient lighting condition, dusty after blasting and low reso-
lution of the captured images [1].

For the algorithm to perform successfully, a large amount of training data is required.This research introduces the
use of drone-generated aerial photography to the fragmentation data collection stage in mining and quarry applica-
tions. Furthermore, the Aerial Drone-captured Images will then undergo a recognition process, where the data is also
used for training and learning based on a machine learning technique, Deep Neural Networks (DNNs), in return
toimprove the performance of computer vision tasks [2]. This technique involves injecting prior knowledge of
analyzed images (training data) into a mathematical model. This model can then be used to analyze new images at
high precision and accuracy.

The drone-captured aerial images will be merged and formatted into 512x512 color image slices using Script label-
ling technique (the boundary is labeled blue; the fines are labelled red and the mask is labelled cyan) [3].

Furthermore, a simple Convolutional Neural Network (CNN) Model is applied to the labelled images to extract size
distribution information. In order to assess the accuracy of the model, the bias (the difference between train loss and
human measurement) and the variance (difference between validation loss and train loss) are recorded assuming
none error in human-level measurements. In the long term, the simple CNN model will be trained with massive data-
set via iteration of previous data and new data targeting at a mature model where the bias and variance are minimized
to an acceptable range. The improved model will widely fit most situations in mine sites. Overall, in tandem with
photo analyses software, the application of drone imaging gives quarry and mine sites a fast, accurate and economi-
cal method in order to achieve an optimal performance of all the components (loading crushing milling and other
heavy operations) to finally meet the product specification.

[1] Rotterdam, Netherland; Brookfield, VT, USA: A.A. Balkema (1996). Measurement of blast fragmentation:
proceedings of the Fragblast-5 Workshop on Measurement of Blast Fragmentation, Montreal, Quebec, Canada,
23-24 August,1996

[2] Mahdi Ramezani, Saman Nouranian, Ian Bell, Bahram Sameti and Shahram Tafazoli Motion Metrics Interna-
tional Corp (2017). Fast Rock Segmentation Using Artificial Intelligence to Approach Human-Level Accuracy,
Canada, 2017

[3] Tamir, R, (2017). Utilization of Aerial Drones to Optimize Blast and Stockpile Fragmentation, Nashville, TN,
USA, 2017. UnERD
2O19 • 72
AUG 13
G2: Applying Fenton's Reagent as An Organic Digestion Method to Aid in Microplastic Extraction
in Stormwater Samples
Ziting (Judy) Xia1, Kelsey Smyth2, Elodie Passeport1,2

Department of Chemical Engineering and Applied Chemistry, University of Toronto; 2Department of Civil & Mineral Engineering,
1

University of Toronto

Microplastics contamination in the marine environment has emerged as a global issue with more than 300 million
tons of debris floating on sea surfaces. Recently, stormwater run-off is hypothesized as a major pathway for
transporting microplastics from land-based sources to the oceans. The Passeport Lab has dedicated a project to
establishing this correlation. Stormwater samples are density separated to allow the microplastics extraction for
characterization. However, organic substrates in these samples mask the density of the microplastic particles
(MPs), resulting in a low extraction yield. This study aims to modify and validate Fenton's reagent, an iron-cata-
lyzed oxidation reaction commonly applied to wastewater treatment, as an additional organic digestion step before
the density separation. The study consists of 2 phases. Phase I ensures Fenton's reagent does minimal damages to
the MPs, while Phase II tests its efficiency in removing organic contents. The study had passed Phase I since the
before-and-after surface areas of the reacted MPs showed no significant changes, further confirmed by spectro-
scopic results. Currently, the study is on Phase II, where two soil replicates are subjected to Fenton's reagent and
standard Loss-on-Ignition respectively. The mass losses of both soil samples post-treatment indicate the organic
mass in soil and will be compared against each other. No significant difference is expected. The findings of Phase
II will immediately impact the microplastic project by increasing the microplastics extraction yield, thus determin-
ing the fate of the MPs in stormwater run-off. Further, this study will contribute to developing a systematic proto-
col for microplastic extraction.

73 • UnERD
2O19
AUG 13
 H: Food and Nutrition
H1: Iron Content in Hibiscus sabdariffa Calyces: Optimization of Extraction Conditions
Diana Teichman1, Ade. O. Oyewole2 and Levente L. Diosady2,3
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto

The World Health Organization estimates that globally 1.62 billion people suffer from anaemia [1]. Iron deficiency
(ID) is a leading cause and is most prevalent in developing countries [2]. ID has a negative impact on the health and
economy of affected populations. Therefore, there is a need to increase iron intakes from staple/commonly
consumed foods and beverages.

The aqueous extract of Hibiscus sabdariffa calyces, a commonly consumed beverage in many developing countries
is reported to be relatively rich in iron [3]. Potentially, this could be useful to combat ID. This study aims to optimize
extraction conditions to obtain the best yield of iron: by determining the optimal temperature, steeping time and
solid to solvent ratio.

Preliminarily, following conventional extraction methods, seven different weights (1.25–25g) of dried calyces were
heated at 90°C in 100ml of water for 30 minutes. Cooled samples were strained, washed and vacuum filtered. Resi-
dues and filtrates were weighed and dried in an oven at 105°C. Dried samples were analyzed using ICP-AES. The
iron content in raw dried calyces was determined to be 9.40±0.43mg/100g and the percentage of extractable iron in
the filtrates was 34.2±6.4%. Based on iron yield and desired extract consistency, 6.25g/100ml and 12.5g/100ml were
determined to be workable solid to solvent ratios for the optimization design of experiments (DOE).
Ongoing experiments (a 23 DOE) will investigate the effect of all factors (solid to solvent ratio, temperature and
steeping time) on iron yield. Hence, optimal extraction conditions giving the highest iron content in the beverage
will be determined.

[1] “World Health Organization: Global anaemia prevalence and number of individuals affected,” Who.int. [On-
line]. Available: https://www.who.int/vmnis/anaemia/prevalence/summary/anaemia_data_status_t2/en. [Accessed:
27- Jul- 2019].

[2] “Anemia - Symptoms and causes,” Mayo Clinic, 2019. [Online]. Available: https://www.mayoclinic.org/dis-
eases-conditions/anemia/symptoms-causes/syc-20351360. [Accessed: 27- Jul- 2019].

[3] I. Da-Costa-Rocha, B. Bonnlaender, H. Sievers, I. Pischel and M. Heinrich, “Hibiscus sabdariffa L. – A

phytochemical and pharmacological review,” Food Chemistry, vol. 165, pp. 424-443, 2014. Available: 10.1016/j.-
foodchem.2014.05.002.

UnERD
2O19 • 74
AUG 13
H2: Determination of Micronutrients Fortification Capacity in Bouillon Cubes Using Proximate
Analyses
Gladys Olubowale, Kiki Chan, Levente Diosady
1
Department of Chemical Engineering and Applied Chemistry, University of Toronto.

Micronutrient deficiencies affect over two billion people in the world today due to insufficient intake of micronu-
trients such as iron, iodine and vitamin A. To help improve consumer health, particularly in low- and middle-in-
come countries, daily intake of these micronutrients could be increased by fortifying staple foods. There is ongo-
ing research on identifying appropriate food vehicles to be fortified. Fortification efforts on salt, tea and cooking
oil provide guidelines on the selection of food vehicles as well as suitable micronutrients to be added. Bouillon
cubes have emerged as a potential candidate for food fortification. Bouillon cubes are seasoning ingredient com-
posed of salt, fat, proteins, starch, herbs, spices, flavourings. Bouillons cubes are widely consumed in households
across all socioeconomic levels in Central and West Africa at 1-4g per person per day. Frequent consumption
coupled with low cost and self-limiting dosage makes bouillon cubes a promising vehicle for micronutrient fortifi-
cation. However, little is known about the fortification capacity of bouillon cubes. To examine the fortification
capacity of bouillon cubes, proximate analyses were conducted. Proximate analyses determine the amount of fat,
moisture, protein, carbohydrate, fibre and ash in a given food item. The proximate composition of three brands of
chicken bouillon cubes were analysed. Data acquired from each brand were used to determine the composition of
bouillon cubes that are commercially available. These results helps to place the current nutrient content of avail-
able bouillon cubes in perspective and hence provide the background for further decisions on the way forward in
terms of the needed composition to make fortified bouillon cubes.

75 • UnERD
2O19
AUG 13
 ACKNOWLEDGEMENTS
SPONSORS & PARTNERS

JUDGES
IBBME Mr. Leo Zhu Ms. Georgiana Moldoveanu Prof. Natalie E. Jerger

Prof. Alison McGuigan UTIAS Mr. Randeep Gabhi Electrical

Prof. Christopher M. Yip Prof. M. Reza Emami Ms. Courtney Toth Prof. Peter R. Herman

Prof. Edmond Young Mr. Matthew Giamou Dr. Ashraf Amin Prof. Mo Mojahedi

Dr. Muhammad Rizwan Prof. James Davis Dr. Jayesh Sonawan Prof. Wai Tung Ng

Ms. Nataly Machado Siqueira Mech MSE Prof. Khoman P -hang

Dr. Simon Pascual Gil de Gomez Prof. Sanjeev Chandra Prof. Thomas W. Coyle Mr. Qingnan Yu

Mr. Rob Liddell Prof. Eric Diller Prof. Chandra Veer Singh

Mr. Wilson Dr. Pantea Tavakolian Ms. Jeanne N'Diaye

ECP Mr. Guorui Wang Indy

Prof. Ken Tallman Mr. Mahdi Hamidi Prof. Mark H. Chignell

Prof. Alan Chong Dr. Vincent Iacobellis Prof. Mark S. Fox

Prof. Robert Irish Chemical Prof. Viliam Makis

Prof. Deborah Tihanyi Prof. Arthur Chan Mr. Aloagbaye Momodu

Prof. Peter Weiss Prof. Greg J. Evans Computer

Dr. Lydia Wilkinson Mr. Daniel Di Matteo Prof. Raviraj Adve

Mr. Shervin Mehryar Prof. Elodie Passeport Prof. Jason H. Anderson

UNERD 2019 PLANNING COMITTEE

Laura Berneaga, Jennifer Wang (Co-Chairs) My My Tran (Marketing)

Shevien Rhule (Faculty Liason) Muskan Sethi (Sponsorship and Awards)

Hilary Fung (Finance) Maeesha Biswas, Jiazhen Li (Programming)

Zizhao (Zoe) Chen, Shauda Islam (Judging) Thandi Myers, Safwan Hossain (Logistics)

FACULTY & STAFF ADVISORS

Cori Hanson (Contact & Support Staff) Shevien Rhule (Faculty Liaison) UnERD
2O19 • 76
AUG 13
Notes:
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
Notes:
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
____________________________________________________
UnERD
2O19
AUG 13

booklet design by My My Tran