University of Southern California Dissertations and Theses

3D printing and compression testing of biomimetic structures

(USC Thesis Other)

3D printing and compression testing of biomimetic structures

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 3D Printing and Compression Testing
of Biomimetic Structures

By
Lian Lash-Rosenberg
________________________________________________________________________

A Thesis Presented to the
Faculty of the USC Viterbi School of Engineering
University of Southern California

In partial fulfillment of the
Requirements for the Degree
Master of Science
(Mechanical Engineering)

May 2018

Copyright 2018 Lian Lash-Rosenberg
2
Acknowledgements
Throughout the process of writing my thesis and conducting research there was
much ground to cover and I would not have completed it without the help and support of
many people. First and foremost, I would like to thank Dr. Andrea Armani for giving me
the opportunity to work in the Armani Research Lab and supporting me on a variety of
different projects. I met Dr. Armani during the fall semester of my freshman year at USC
and throughout the years I have greatly appreciated her encouragement, assistance, and
her willingness to take the time to talk about lab, classes, work and future directions.
Having a professor as accessible and caring as Dr. Armani was crucial to my success
during my time at USC.
Many thanks to Dr. Michael Kassner and Dr. Ivan Bermejo-Moreno for their
support and service on my thesis committee. I would also like to thank Dr. Qiming Wang
for letting me use his lab facilities and providing me with invaluable resources. I greatly
appreciate Kun Hao Yu for sharing his knowledge on the 3D printing process. Learning
from someone who has as much familiarity with the methods as he has, made fabricating
the 3D printed structures both interesting and fun.
I would also like to thank Alexa Hudnut for helping me with the testing of
samples and for mentoring me throughout my time in graduate school. I am very
thankful for the one-on-one meetings, as well as ensuring that I remained on track and
understood how my research fits in with other projects in the lab. I am also grateful for
the proofreading of my paper and her many insightful comments and questions. My time
3
in lab was made thoroughly enjoyable by all of the people in it and the supportive, fun
environment they brought to work every day.
I would also like to thank my friends and family for supporting me through this
process. Thank you to Charissa Seid for working alongside me and helping me stay
focused. Thank you to Kevin Kim for bringing me snacks and keeping me motivated. In
no particular order, I greatly appreciate Connor Kerns, Devin Rousso, Megan Hansford,
Cole Feagler, Julia Roche, and Brendan Dalbow for giving me feedback on this paper.
All of your comments and questions were insightful and extremely helpful in making
sure that this paper was clear and concise. Connor Kerns, thank you for ensuring that I
was thorough in my background explanation for those without extensive biology
knowledge and for keeping me laughing with your clarifications. Devin Rousso, thank
you for reading my paper multiple times and catching my repeated use of the same set of
words. Megan Hansford, thank you for letting me use your apartment as a workspace,
keeping me company while I wrote, and for making me breakfast and coffee. Cole
Feagler, thank you for being as excited as I am about the research that I am doing. Julia
Roche, thank you for encouraging me throughout the weeks leading up to my completion
of this paper and for being the first person other than myself to read it through. Brendan
Dalbow, thank you for being encouraging and motivational; I really appreciate all support
and effort you put into helping me succeed. I am also thankful to Colleen Motoyasu,
Mahrukh Fatima, David Ruiz, Vinh Nguyen, Brennan Graham, Carl Salazar, Chris
Casalaspi, Chris Luong, and Marissa Gustavson for feedback on my defense presentation.
I am very grateful to my family for supporting me throughout my higher education and
teaching me the value of hard work and learning.
4
I would not be completing my thesis without all of you and for that I am
extremely thankful.
5
Table of Contents
Acknowledgements ..................................................................................................................... 2
Table of Contents ........................................................................................................................ 5
List of Figures .............................................................................................................................. 6
Abstract ......................................................................................................................................... 7
Chapter 1: Background ............................................................................................................. 9
1.1 Mechanical Properties .................................................................................................................. 9
1.2 Mechanical Study of Tissue Samples ...................................................................................... 11
1.3 Mechanical Testing of Biological Samples ............................................................................ 16
1.4 Biomimetic Models for Treatment Testing ........................................................................... 21
1.5 Three-Dimensional Printing ..................................................................................................... 23
Chapter 2: 3D Printing Biomimetic Structures ................................................................ 30
2.1 Structure Design .......................................................................................................................... 30
2.2 Structure Modeling ..................................................................................................................... 32
2.3 Structure Fabrication ................................................................................................................. 34
2.4 3D Printing Challenges .............................................................................................................. 42
Chapter 3: Mechanical Characterization .......................................................................... 44
3.1 Sample Testing ............................................................................................................................. 44
3.2 Data Analysis ................................................................................................................................ 46
Chapter 4: Future Directions and Conclusions ................................................................ 53
4.1 Conclusions ................................................................................................................................... 53
4.2 Future Directions ........................................................................................................................ 53
References .................................................................................................................................. 57

6
List of Figures

Figure 1: Tensile and compressive testing .......................................................................... 9
Figure 2: Stress-Strain curve ............................................................................................. 10
Figure 3: Liver lobule ....................................................................................................... 13
Figure 4: Kidney ............................................................................................................... 14
Figure 5: Optical fiber photoelastic polarimetry ............................................................... 21
Figure 6: Example of extrusion based 3D printing ........................................................... 24
Figure 7: Comparison between structures with and without supports .............................. 25
Figure 8: Projection microstereolithography .................................................................... 26
Figure 9: Projection microstereolithography; dual material printer ................................. 27
Figure 10: Rendering of six different lattice structures .................................................... 31
Figure 11: Steps for modeling the 3D printed structures .................................................. 33
Figure 12: Rendering of 3D printed structures ................................................................. 34
Figure 13: 3D printing process ......................................................................................... 35
Figure 14: Run sheet used to create 3D printed parts ....................................................... 36
Figure 15: Example image layers for 3D printing ........................................................... 38
Figure 16: Shows the steps of silicone injection ............................................................... 40
Figure 17: Dissolving shells.............................................................................................. 41
Figure 18: Picture of Instron compression testing setup with sample. ............................. 45
Figure 19: Sample indexing .............................................................................................. 46
Figure 20: Compression curves at 10%, 20%, and 30% strain ......................................... 47
Figure 21: Compression of Sample 1 using Instron machine ........................................... 49
Figure 22: Study of Sample 1 ........................................................................................... 50
Figure 23: Minimal variance between runs ...................................................................... 51
Figure 24: Buckling .......................................................................................................... 52
Figure 25: Testing setup using Instron and Optical Fiber Polarimetric Elastography ...... 54

7
Abstract
The following thesis outlines the results of a series of studies on the mechanical
properties of tissue samples in comparison to three dimensional (3D) printed silicone
structures. The mechanical behavior of tissue is hard to study due to the lack of
standardization of testing methods. Tissue also undergoes physical changes after
extraction, which alter the properties of the samples. Accurate modeling of tissue
samples has the potential to provide an easier path to greater understanding of tissue
structure. This would lead to more accurate disease modeling through benchtop testing
of biomimetic samples, resulting in more enhanced treatments.
The primary 3D printing process utilized during this project is called projection
microstereolithography. This is used to create a hollow shell of the simplified, modeled
tissue structure. Silicone is then injected into the shell, which, in turn, is dissolved away
once the silicone has cured. This leaves the finished structure, which is ready for
compression testing.
Testing the compressive properties of tissue yields significant information about
the sample. The goal of this project was to be able to mimic these tissue tests with
corresponding fabricated samples. This was achieved by compressing six varied
structures. The first was the base model, comprised of a patterned XX structure, which
was modified to create varied samples with induced defects. From the initial model,
other structures were created by removing either one or two trusses in varying locations.
The samples were then tested at three different strains, 10%, 20%, and 30%, to look at the
effect of different load conditions on the buckling point of the sample.
8
Being able to accurately reflect the properties of various tissues using synthetic
models will help standardize the mechanical testing of biological material and provide an
accurate structure for future biological study. Synthetic models would eliminate the time
constraints of tissue testing, as well as the sensitivity to contamination. They could also
provide cheaper models for testing the effect of various treatments on the body.
Furthermore, additional coordination between biological and mechanical technology
would aid in the advancement of tissue research.

9
Chapter 1: Background
1.1 Mechanical Properties
Mechanical properties are a good method for classifying different materials. They
are predictive of how a sample will respond to outside forces based on the intrinsic
material properties, as well as the shape of the material. These are based on experimental
results from varying tests. Tensile or compressive testing is the most common and
informative. Tensile testing pulls the ends of the sample in opposite directions while
compression testing pushes the ends of the sample together (Fung 1981).

Figure 1: Tensile and compressive testing
a) Shows sample, in orange, under tensile load. Sample is pulled apart resulting in
permanent, plastic deformation. b) Shows sample under compressive load. Sample
is compressed until bucking occurs.

Materials have a region of both elastic and plastic deformation. If the load is
removed during the region of elastic deformation, the sample will return to its original
a) b)
10
state with no lasting effects from the load. If the load is applied past the linearly elastic
region, the sample becomes plastically deformed and undergoes permanent change. At
this point, if the load is removed, the sample will not have the same initial properties
(Fung 1981). Figure 2 shows a graph of the stress applied to the sample verses the
material’s strain response during the tensile testing of a part.

Figure 2: Stress-Strain curve
The graph shows the relationship between the applied stress and the material strain
on a part during tensile testing. Under minimal loads, the part experiences elastic
deformation, indicated by the linear portion of the graph before the green x. After
this point, the part undergoes plastic deformation, permanently altering the part.
The red x represents fracture, when the material breaks as a result of the increased
load.

The stress is calculated as force per unit area, which looks at how much load is
being applied to a section of material. This is measured in units of Newton per square
meter, or Pascals (Pa). This is graphed against the strain, which is unitless. The strain is
x
Strain (ɛ)
Stress (σ)
x
11
calculated by dividing the change in length by the original length of the sample. As the
sample is put under increasing tensile load, it will stretch more, thus increasing the
change in length.
While tensile testing looks at the point of fracture, compressive load focuses on
the buckling point; the point at which the material loses its load carrying ability. In
tensile testing, once the material fractures, it is broken into two pieces and no more data
can be gleaned. With compressive testing, load can still be applied to the material after
buckling, but it is difficult to study given its unpredictable nature. Everything that occurs
after the buckling point of the material is called the hysteresis, which focuses on the
effect of past testing on the material properties. This presents difficulty in studying the
repeatability of experiments given that one test has the ability to impact later runs (Fung
1981).
Mechanical properties are studied in all materials, however they are especially
important with regards to biological material. The body is under the influence of outside
stimuli all the time; therefore it is important to be able to predict how different tissues
will respond in varying situations. Furthermore, how the body reacts after a change in
mechanical forces is highly relevant to healing, whether that be from a disease or general
aging (Fung 1981).

1.2 Mechanical Study of Tissue Samples
The compression testing method was used to compare and study three main types
of tissue: liver, pancreas, and kidney. While all tissues degrade, the pancreas breaks
down exceptionally fast due to the large quantities of enzymes that are used in digestion,
12
predominantly for nutrient absorption. The pancreas also plays a key role in the
endocrine system, creating and secreting hormones such as insulin and glucagon (Kasper,
Fauci et al. 2015). The pancreas is especially relevant to the study of mechanical
characterizations because the complexity of the pancreas can be largely accredited to its
structure. The pancreas is composed of two types of tissue; exocrine and endocrine.
Each of these two structures consists of small clusters of cells that form larger structures.
More specifically, the pancreas consists primarily of exocrine tissue, which is comprised
of many small ducts that join together to form large ducts. These ducts are made up of
acinar cells, which emit digestive enzymes. Endocrine tissue is composed of small
bundles of cells connected by a network of capillaries (Kasper, Fauci et al. 2015). These
intricate structures make modeling the pancreas especially difficult but all the more
crucial to fully understanding the organ.
Like the pancreas, the liver also aids in digestion. The liver’s main function is to
filter blood before it passes from the digestive tract to the rest of the body. It helps
digest, process, and absorb nutrients while managing the amount of sugar in the blood
stream. The liver also produces bile, which helps break down food and metabolize drugs
and other chemicals before they are released back into the intestines. The liver is made
up of four lobes, each comprised of small, hexagonal units called lobules that consist of a
central vein with surrounding portal veins and six hepatic arteries, as seen in Figure 3.
The hepatic arteries are connected to the central vein via sinusoids, which carry two cell
types: Kupffer cells and hepatocytes. Kupffer cells are macrophages that engulf and
digest aged red blood cells. Hepatocytes make up the epithelial cells of the liver and help
regulate osmotic pressure by synthesizing albumin, the main protein in blood plasma.
13
The albumin is used to transport water, cations, hormones, and fatty acids. The liver is
structurally reinforced by the peritoneum, which is the membrane that lines the cavity of
the abdomen (Jain, Damania et al. 2014, Kasper, Fauci et al. 2015). Given the structural
complexity of the liver, being able to model its unique mechanical structure is crucial to
the progression of understanding the liver.

Figure 3: Liver lobule
Hexagonal function unit of the liver
(Jain, Damania et al. 2014, Kasper, Fauci et al. 2015)

The final organ studied was the kidney. Kidneys are key in the filtration and
reabsorption of nutrients before waste exits the body. They are made up of nephrons,
which are composed of complex, tubular structures that allow water and nutrients to be
reabsorbed into the body. As fluid flows through these tubes, waste is collected and exits
the body as urine. The tubes are permeable to different substances, such as water and
Central Vein
Portal Triad
Hepatocytes
14
salts, to ensure that there is an optimal concentration of different components inside the
body. Each kidney contains roughly one million nephrons, all of which comprise an
entire filtration system ( Kasper, Fauci et al. 2015, Kurts, Panzer et al. 2013).

Figure 4: Kidney
a) Rendering of a kidney. The red tubes indicate unfiltered blood going into the
kidney, while the blue tubes indicated filtered blood leaving the kidney.
b) Rendering of a nephron, the functional unit of the kidney ( Kasper, Fauci
et al. 2015, Kurts, Panzer et al. 2013).

These three tissues were chosen due to their accessibility and consistent results;
both important characteristics for testing the feasibility of these methods. Of the three,
the pancreas is particularly challenging. Due to its participation in both the endocrine
and exocrine system, it contains an abundance of enzymes, which cause rapid
degradation upon removal from the host. Furthermore, the body has a unique reaction to
pancreatic cancer; it will naturally isolate cancerous tumor cells, creating a section of
tissue with higher stiffness (Cross, Jin et al. 2007, Kasper, Fauci et al. 2015). This makes
a) b)
15
it an interesting target to study, with distinct application to the changing mechanical
properties of tissue.
Creating accurate models of these intricate structures is not easy. Given the
complexity of the functional unit of each organ, being able to model a scalable, easily
reproducible unit cell would significantly aid the study of the response of these organs to
potential stimuli and treatments. Scalability is crucial because so much of tissue
biomechanics is due to the ways in which cells are connected to each other by the
extracellular matrix (ECM). Furthermore, due to the intricacy and importance of each
unit to the function of the organ as a whole, study of these units could improve larger
models and benchtop testing of disease treatments (Zhang, Montgomery et al. 2016).
In addition to the challenges stemming from attempting to recreate their highly
intricate structures, testing the tissues themselves presents many difficulties. Rigor
mortis begins to set in not long after the tissue is separated from the host, causing drastic
effects on the mechanical properties of the tissue shortly after removal. Rigor mortis is
the process of the tissue stiffening as the supply of adenosine triphosphate is depleted.
Adenosine triphosphate is produced by the body, predominantly through respiration, and
is used in the process of relaxing muscles. Once the tissue is cut out of the host and no
longer has access to oxygen to create adeno1sine triphosphate, the relaxation of muscles
is prevented and the tissue becomes rigid (Bonzon, Schön et al. 2015, Kasper, Fauci et al.
2015). This stiffness changes many aspects of the mechanical properties of the tissue,
one of which is how the tissue responds to compression.
Given the challenges of studying biological tissue samples, creating accurate
models would greatly improve this process. Having a standardized testing setup for
16
extracted tissue samples would also improve these studies. A combination of utilizing
biomimetic samples as well as nondestructive, portable testing methods would increase
testing efficiency and accuracy, thus allowing a more comprehensive understanding of
biological tissue. The advances in 3D printing that were mentioned previously provide a
method for achieving these goals. By 3D printing structures that mimic the ECM of
specific organs, we aim to measure the impact of geometric changes on biomechanics. In
turn, this will enable a better understanding of disease progression by providing a
framework to understand how changes in biomechanics directly correlate to structural
changes within the ECM.

1.3 Mechanical Testing of Biological Samples
Although difficult to measure and sometimes overlooked, the tissue’s response to
mechanical stresses is a key facet of how the tissue functions. Many ailments are
reflected in the mechanical properties of the tissue. For example, a heart attack greatly
affects the mechanical properties of the impacted area. The tissue hardens, thus creating
scar tissue that is less elastic. This is somewhat constricting to the heart, which requires
elasticity to pump blood throughout the body (Fung 1981, Kasper, Fauci et al. 2015).
Cartilage impact is another key example to the importance of tissue properties. As time
passes, the bone continuously wears down the cartilage. This causes the cartilage to
degrade, which eventually results in bone on bone contact. The mechanical compression
of the cartilage is key to preventing this direct impact (Radin and Paul 1971, Fung 1981).
Furthermore, the ECM in cancerous tissue, stiffens while the cells become more flexible
(Fung 1981, Cross, Jin et al. 2007, Kasper, Fauci et al. 2015). In addition to the
17
detrimental effects of cancer, the mechanical properties of the tissue are altered and
therefore do not exhibit their natural response to mechanical load.
One example of how the biomechanics of tissue impact biological function is that
the differentiation of stem cells due to the geometry of the ECM scaffolding. The ECM
helps provide the correct stimuli, causing the stem cells to exhibit different cellular
properties based on the surrounding environment (Kharaziha, Shin et al. 2014, Guo, Yu
et al. 2016). Given this, studying the mechanical properties of the tissue and being able
to accurately represent the extracellular matrix will help improve researchers’
understanding and ability to mimic tissue in their tissue engineering efforts. Being able
to use synthesized material to replace damaged components, such as worn cartilage has
many applications. Over time, cartilage in the knees wears down, reducing the padding
between the bones thereby making it extremely painful to walk (Radin and Paul 1971,
Fung 1981). If this cushion could be replicated and added between the bones, the
necessity of joint replacements could be eliminated and the quality of life for many
people would be greatly improved.
An important element of understanding tissue is having accurate models from
which to learn. One of the key issues in the study of biological tissue is the difference
between testing an extracted sample and testing the host, where the tissue is in its natural
environment. Excised tissue often responds differently than how it would inside the
body. Furthermore most samples accurately reflect a native response for only a limited
period of time after extraction, making testing quite difficult due to time sensitivity (Bate-
Smith and Bendall 1949). In addition to rigor mortis causing the mechanical properties
of the tissue to change, the process of autolysis also alters the tissue samples as the cell
18
enzymes begin to degrade themselves (Kasper, Fauci et al. 2015). This limits where
experiments can be performed, as well as how much time the researcher has to complete
their testing. In many situations, the tissue sample has to be extracted from a patient in
the operating room and then transported to a research lab where testing can be conducted;
a process that can take up to an hour. Given that the transport and the testing both take
time, immediate experimentation after tissue removal is crucial. Once the tissue is
removed from its host, the enzymes within the sample begin to degrade the surrounding
tissue. Over time, this greatly alters its mechanical characteristics, which eventually
causes skewed results that are not consistent with a natural response (Bate-Smith and
Bendall 1949).
This is an exemplary model for addressing the increasing concerns about
reproducibility of results within academic and industry research (Collins and Tabak 2014,
Baker 2016). One way to address these concerns is to model tissue samples with 3D
printing. Using this method to create models out of silicone would solve both the time
sensitivity and the varied behavior of the sample. Unlike tissue, silicone does not change
over time so there is no time constraint on the testing. Furthermore, the geometry of
silicone is easily manipulated; the silicone can be mixed with a catalyst in liquid form
and injected into a mold. Additionally, silicone is not a biohazard, which allows it to be
tested on the same fixtures as tissue samples (Jiang and Wang 2016).
Efforts have been made to attempt to measure the mechanical properties of tissue
using several different methods such as Atomic Force Microscopy and Rheology. While
these methods are somewhat effective, they lack the resolution and flexibility required for
tissue testing. Their design does not account for the heterogeneity of tissue or the time
19
constraints associated with tissue testing (Cross, Jin et al. 2007, Ayyildiz, Cinoglu et al.
2015). A more accurate testing set up would have higher resolution in differentiating
between various sections of the sample rather than obtaining the properties of the entire
tissue section. Furthermore and ideal method would also reduce the time between tissue
resection and testing.
While the testing of the tissue samples themselves present many challenges, the
experiments also frequently damage the sample. Compression testing is a particularly
destructive test because the extra cellular matrix, which determines most of the
mechanical properties of the tissue, may not be accustomed to heavy strain. For example,
during daily life, bones experience a lot of compression. This trains the cells to be used
to this type of loading. However, if a sample of heart tissue, were placed under the same
type of loading, there would be severe damage to the cardiac myocytes (Bose,
Vahabzadeh et al. 2013). This compressive damage often prevents the samples from
being retested, therefore making repeatability difficult to obtain. Methods for non-
destructive testing have only recently become viable (Hudnut, Babaei et al. 2017). These
newly developed methods are based on a non-destructive optical fiber photoelastic
polarimetry system (OFPE) that can be used to preserve the intrinsic tissue structure even
after compression. Researchers compared histology images from liver, kidney, and
pancreas tissue samples before and after compression testing. Surprisingly, they found
the cellar borders and nuclei were undamaged when testing was performed using this new
method (Hudnut, Babaei et al. 2017). This is substantial, given that compression testing
of tissue reflects important sample properties. The elasticity of tissue has been found to
span up to eight orders of magnitude, from 50 Pa all the way up to 5 GPa, 5x10
9
Pa
20
(Hudnut, Babaei et al. 2017). From this, it is clear that the compression testing of tissue
samples is vital to determining the ideal biomimetic structure.
Optical fiber photoelastic polarimetry is well suited for testing the compressive
properties of tissue because it is highly tunable and provides impressive resolution
(Harrison and Armani 2015, Hudnut and Armani 2017, Hudnut, Babaei et al. 2017).
When performing this test, an optical fiber is placed underneath the sample. This can be
seen in Figure 5, represented by the thin blue line. One end of the fiber is connected to a
laser light source, the green box in Figure 5, while the other end feeds into a specialized
sensor called a polarization detector, the red box in Figure 5. The sample is then
compressed with an automated microcontroller. As a result of the force on the optical
fiber, the refractive index of the fiber changes; a phenomenon called the photoelastic
effect. The variation in the refractive index, in turn, alters the polarization of the optical
field. The change in the polarization state can then be read by the polarization detector.
This method provides a more detailed depiction of the material response to compression
because the system only responds to forces applied to the section of tissue directly above
the optical fiber (Harrison and Armani 2015, Hudnut and Armani 2017, Hudnut, Babaei
et al. 2017). This allows researchers to observe the tissue response to compression in
great detail because it focuses on one section of the sample, rather than the entire surface.
This is ideal for nonhomogeneous samples because it allows for the comparison of
different sections of the samples.
21

Figure 5: Optical fiber photoelastic polarimetry
Diagram of compression system. The orange square indicates the location of the
sample to be tested. The blue bar compresses the sample, pushing down on it in
increments of force dictated by a micrometer (Hudnut, Babaei et al. 2017).

Furthermore, the testing setup is small, portable, and has disposable components
(Harrison and Armani 2015, Hudnut, Babaei et al. 2017). Disposability makes it
convenient for testing biological samples because contamination is a common issue,
while portability enables the testing of tissue samples at the site of extraction, which
drastically cuts down testing time, thus preserving the integrity of the samples.

1.4 Biomimetic Models for Treatment Testing
One of the primary areas of research within the field of tissue engineering
involves the search for, and development of, biomimetic scaffolding to populate with
stem cells. Initially, even achieving mechanical accuracy with scaffolding models would
be a large step forward. While researchers have succeeded in using decellularized organs
to provide the ideal natural structure, the demand for organ transplants is still greater than
Polarimeter
Laser
Micrometer
controlled
compressor
Sample
Optical Fiber
Bare Fiber
Adapter
22
the supply of donors (Harald, Thomas et al. 2008). Being able to use 3D printing to
create structures that can accurately mimic the mechanical properties of these organs
would help the progression of this field toward being able to 3D print the organs
themselves, thus being able to meet the high demand for organ transplants.
Being able to use 3D printing to create accurate biomimetic models, not only for
mechanical characteristics, but also for other physiological traits as well, would be
beneficial for testing the mechanical properties of tissue and could be used to better
simulate the body’s response to various treatments. Although this is currently not
possible, it would be highly advantageous. The pharmaceutical industry offers an
excellent example of this process; before drugs are released to the market, they are
heavily tested through in vitro benchtop testing and in vivo clinical studies. Having
structurally accurate models to use for benchtop testing would provide more precise
results on how the drug would react inside a host, thus reducing the risk when performing
initial patient studies ( Kang, Lee et al. 2016, Zhang, Montgomery et al. 2016).
Fully understanding the properties of different tissues can also help with disease
modeling. Curing diseases always starts with an extensive understanding of how the
disease functions. To obtain this understanding, an accurate model is key to studying
how one aspect of the body is impacted during disease progression. Similar to drug
testing, being able to have a model of the tissue structure would greatly improve
treatment studies by indicating the tissue ECM response (Zhang, Montgomery et al.
2016).

23
1.5 Three-Dimensional Printing
In the field of biological tissue research, drastic advancements in 3D printing have
made the technologies for fabrication increasingly more applicable. This area of study
has advanced from single material, extrusion-based printing to powder printing, inkjet
printing, and projection microstereolithography (Bose, Vahabzadeh et al. 2013). 3D
printing can be used alongside many different types of computer aided design software.
A model is created, converted to a file that can be read by the printer, and then loaded
onto the 3D printer’s interface. Given the many different types of printing, it is easy to
tailor the process to whatever is being fabricated. For example, with extruder based
printing, the material has a greater likelihood of overflowing or spreading out, which
makes the print less accurate. The extrusion of supporting structures also becomes
necessary if the part has too much overhang. While tolerance may be sacrificed, the
prints are inexpensive and do not require an extensive run time (Hamidi, Jain et al. 2017).
One application of this technology is 3D printing scale models of different organisms.
Figure 6 demonstrates using 3D printing to create a scale model of a Tetraodontidae. The
tail of the Tetraodontidae would not be able to be completed without supports because the
extruder head would have no base to build on. This example demonstrates the current
limitation of extrusion 3D printing methods.
24

Figure 6: Example of extrusion based 3D printing
a) Shows extruder head printing raft and supports for the model.
b) Final 3D printed model with the supports still attached.

Other methods of printing do not require supporting structures because they are
built into how the part is fabricated. Powder printing uses laser-cured powder to create
each layer. This process uses a powder base for every layer, which allows the previous
layer of cured and uncured powder to serve as the supporting material for the build (Bose,
Vahabzadeh et al. 2013). This process is ideal for creating intricate structures on the
order of micrometers, because it does not require structural supports and has high
resolution (Bose, Vahabzadeh et al. 2013). If the same structure was made using
extrusion-based printing, structural supports would need to be removed after printing, as
they are not part of the final model. This would be challenging for a small, highly
detailed structure given that the supports would need to be removed by hand. Extrusion-
based printing predominately prints in one material; therefore the only way to extract the
part from the supports would be to break the weakened connections between the part and
the support structure. On a model like the Tetraodontidae, this is not challenging.
a) b)
25
However, on a part that is on the scale of millimeters or micrometers, removing these
supports would have to be done with tweezers and is subject to user error.

Figure 7: Comparison between structures with and without supports
The blue represents the desired 3D printed structure.
a) Shows the post-print structure using a method that does not require
additional supporting structure. b) The red represents the supporting
structure that is necessary to an extrusion-based print and would need to be
removed post-production.

Although projection microstereolithography sometimes requires supporting
structures, it is similar to powder printing in that it has the resolution necessary to print
small, accurate models. Microstereolithography uses photocurable liquid to create each
build layer. The print is built from the top down, with the print attached upside-down to
a build plate. The print is represented in Figure 8, by the dark orange structure
underneath the build plate. A UV light source, most often a projector, is placed below a
tray of photocurable liquid. In Figure 8, the uncured liquid is shown in light orange in
the solution well. The projector is synced to a computer program that controls what
images are displayed on the projector screen. The liquid cures and hardens in the
locations that are exposed to the UV light as the build plate moves progressively further
a) b)
26
and further out of the solution. The Z stage motor moves the build plate down into the
solution well to cure each layer of the print. While the build plate is flush with the base
of the solution well, the projector displays a layer of the build, thus curing the solution
that is exposed to UV light (Wang, Jackson et al. 2016).

Figure 8: Projection microstereolithography
The Z stage motor moves the build plate down into the solution well to create each
layer. The base of the delivery wheel has a cutout for the solution well to fit into.
The solution well has a transparent base and is filled with photocurable liquid. The
projector below the solution well displays each image layer, thereby curing the
liquid onto the bottom of the base plate (Jiang and Wang 2016).

This method of 3D printing has many applications due to its high resolution and
tunability. Using stereolithography, several materials can be printed into a cohesive
structure simultaneously. Microstereolithography uses a photopolymerization-based
multimaterial stereolithography system that can print many different materials. A
Projector

Delivery
Wheel
Z Stage Motor
Build Plate
Solution Well
27
delivery wheel that provides the photocurable liquid to the printer can be loaded with two
light sensitive materials, thus allowing it to alternate what material is being printed for
each layer. Figure 9 shows the two-well wheel that allows for a multi-material print. The
delivery wheel rotates 180 degrees after each layer, thus allowing alternating slices to be
made out of different materials if desired. The result is a structure that does not need
supports, is fully fused, and does not require assembly post-production (Wang, Jackson et
al. 2016).

Figure 9: Projection microstereolithography; dual material printer
Diagram of dual material printing system. The computer sends consecutive slices to
the projector, which projects the image onto the base of the solution well. The Z
stage motor moves the glass slide build plate into and out of the well. The delivery
wheel spins 180 degrees between each layer so that a different material is cured onto
the glass slide build plate (Wang, Jackson et al. 2016).

The ability to print two different materials simultaneously has many applications.
Microstereolithography has been used to recreate negative thermal expansion, which is
Projector

Delivery
Wheel
Material Deposition
Z Stage Motor
Build
Plate
Solution Well Solution Well
28
the effect of compression upon the heating of a solid. This phenomenon is most
commonly seen with ice. Unlike most materials, ice decreases in volume when heated to
become water. Being able to fabricate detailed, multimaterial structures has allowed
researchers to study the effects of testing a structure made from two significantly
different materials with drastically different thermal expansion coefficients. This makes
these models very useful for studying negative thermal expansion. The fine-tuning of
this property through varied thermal expansion coefficients yields knowledge that is
especially useful in fields where devices are exposed to varied temperatures, such as
dentistry or microchips (Wang, Jackson et al. 2016). Being able to print with multiple
materials also increases the accuracy of potential biological models given that no tissue is
completely homogenous. Dual material printing allows the combination of multiple
properties to be explored, thus creating a better mimic for tissue. No tissue sample is
made up of a single material with one set of properties; it is an organized mixture of
different components, each with their own characteristics (Hudnut, Babaei et al. 2017).
Photocurable liquids could be chosen for their difference in mechanical characterization
instead of thermal expansion coefficient. In this way, the successful combination of
materials could be applied to more accurate modeling of tissue samples.
Although printing with multiple materials has proven successful, printing solely
with elastic material presents other challenges. In printing with elastic materials, some of
the resolution is lost, and the printable architectures are limited. Furthermore, the overall
print time of softer materials is greater because each layer has to cure for a longer period
of time. The solutions also have high viscosity, which makes them difficult to print with
high resolution. Researchers have also encountered issues where the strength of the
29
material is not able to sustain the weight of the part itself, causing the structure to
collapse in the middle of printing (Jiang and Wang 2016). Given these challenges,
scientists have been applying previously tested methods of 3D printing to create softer
structures. The printing process is used to make hollow molds that can later be removed.
This provides a support structure for the preferred material, which can then be injected,
and cured inside of the shell. The mold is then dissolved away to leave the desired,
elastic model (Jiang and Wang 2016). Injecting and curing elastic material inside of a
shell creates an elastic structure without having to address some of the issues associated
with printing soft material.
It is clear that being able to fabricate and comparably test 3D printed structures
can greatly advance the understanding of tissues. Being able to accurately compare
structural models to native tissue would make testing much easier, given that it could be
performed with models instead of extracted tissue samples. Fortunately, the field of 3D
printing has improved enough to be able to repeatedly create high-resolution parts,
making it much easier to apply these processes to biological uses.

30
Chapter 2: 3D Printing Biomimetic Structures
2.1 Structure Design
The goal of this project was to successfully print and compression test 3D printed
structures. The various structures were made from a base with induced defects using
projection microstereolithography. Before using the 3D printing process, models had to
be designed and created. By mimicking the extracellular matrix of living tissue, 3D
printed models can take many forms. Deciding what base structure to use presented
many challenges given the infinite possible variations in structure design. The design
needed to be easily compressible with a clear buckling point. The lattice structures
presented some difficulty because there are many possible base structures with many
different defects, such as the removal of a truss, or node. The design chosen was easy to
fabricate and induce a variety of structural defects. Being comprised of an ‘X’ shape
made studying the effects of compression on buckling more visually apparent. Tracking
the induced defects in an orderly manner, so as to categorize potential differences
between models, was also integral to ensuring that an entire set of options was covered.
To start, an initial base structure was designed without any defects, specifically chosen
for its simplicity and ease of modifications. The base structure was created with many
planes of symmetry to lower the possibility for varying defects. This meant that, for a
given defect, all permutations could be tested. Aiming to mimic the unknown structure
of tissue, varying defects were induced to see if any comparisons could be made between
compression testing results from porcine pancreatic testing (Hudnut, Babaei et al. 2017).
By removing one, two, or all three trusses from the outside row in varying locations, five
31
additional samples were modeled. Systematically removing trusses, as well as changing
the removal locations, covered all possible combinations for variation in that row. Since
the parts were also rotationally and reflectively symmetric, these truss removals would
yield the same results even if on a different edge. This meant that, for this base, there
were fewer possible combinations of induced defects. Figure 10 shows the removal
locations of the trusses as well as the initial base structure.

Figure 10: Rendering of six different lattice structures
Grey trusses indicate the location where a defect was induced in the form of a truss
removal.

Once the base structure was decided upon, the models were fabricated. 3D
printing was ideal for the characterization of tissue samples because it is easily
customizable with fine details. 3D printing relies on computer aided design (CAD)
models that can be easily manipulated and altered. This makes creating multiple
a) b) c)
d) e) f)
32
iterations of the same structure with controlled defects remarkably easy, a feature that
was highly desirable for this testing. Using CAD software ensured that the base structure
remained the same with the exception of the added defects. This greatly enhanced the
ability to study how each defect affected the compression of that sample. An added
benefit of this method was that these same structures could be imported into Abqus Finite
Element Analysis and modeled computationally.

2.2 Structure Modeling
These structures were modeled in SolidWorks by first drawing the silicone
injection plate. After this, the initial side was drawn and fashioned into two Xs with a set
diameter circle that was used for the cross section of the trusses. The Xs served as the
path of extrusion before being patterned three times to make each row. This can be seen
in Figure 11a and 11b. The same Xs were then drawn again on a perpendicular plane
intersecting the original drawing. The same set diameter was used for the second set of
Xs and the sweep function was used to fill in the trusses that can be seen in Figure 11c.
These were then patterned three times to create the structure that is rendered in Figure
11d. This is the final part, not including the silicone injection plate that is later dissolved
away.
33

Figure 11: Steps for modeling the 3D printed structures
a) Silicone injection plate and initial two X’s. b) Initial X’s patterned three times
across the injection plate. c) Second set of two X’s on a perpendicular plane. d)
Perpendicular X’s patterned three times along the injection plate.

Once the lattice was created, the whole part was shelled to make it hollow. The
thickness of the shell and the diameter of the truss dictated the diameter of the final
structure. Since all of the models were scaled at different points in the process of
fabrication, the shell thickness was made to be 7.5% of the total diameter. Tunnels at the
base of the part were added so that the air and excess silicone inside the tubes could
escape during the injection process. After this final addition, shown in Figure 12b, the
part was deemed ready for printing.

a) b)
c) d)
34

Figure 12: Rendering of 3D printed structures
a) CAD model of the structure before printing modifications were added.
b) CAD model of the structure after the necessary modifications to allow for easy
printing and injection of silicone were added.

2.3 Structure Fabrication
While there are many different methods for 3D printing, the process used for
these experiments is called projection microstereolithography, which utilizes
photocurable liquid to create each layer. Figure 13 shows a picture of the 3D printer used
to create these parts. The projector is used to cure and hardens the liquid on the build
plate in the locations that are exposed to light. The process repeats itself, layer-by-layer
until the whole structure is built. Throughout the entire print, the part itself is suspended
upside-down on a glass slide build plate.
a) b)
35

Figure 13: 3D printing process
The right shows the mechanism that holds the stand and moves it in and out of the
liquid.

To print a structure, a set of spliced images and a run sheet are loaded into
LabView, which is the program that is connected to the printer. The run sheet, displayed
in Figure 14, dictates the parameters of the print, such as the time per layer that the build
plate spends against the base of the solution well and the desired thickness of the build
Stepper
Motor
Stand
Solution
well
Projector
Glass slide
build plate
Part
Membrane
36
layer. This dictated the overall run time of the part, which was roughly two hours. Each
of the rows of the run sheet corresponded to a build layer. The thickness of each layer
was determined by the how far down the build plate was moved relative to the base of the
solution well each time a new layer was created. Figure 14 shows the first 25 layers,
though the indexes continue to 151. The first layer was the base plate, which was thicker
than the other layers because it needed to stick to the glass slide build plate in order to
ensure a successful build.

Figure 14: Run sheet used to create 3D printed parts
The first column indicates the index of the layer. The second column dictates the
layer thickness while the third column specifies the time spent to cure that given
layer. The fourth column indicates the amount of time between layers.

37
Additional software, called Creation Workshop, was used to scale the CAD model
and create the layer images. The model was rotated in Creation Workshop to ensure that
the plate for silicone injection was at the top of the model. Even though the silicone
injection plate is at the top of a model, it is the layer that is the farthest from the glass
slide so that it is the last layer to be printed. This provides holes in the top of the print
that are used for silicone injection once the print is finished. As indicated by the run
sheet, the first layer printed was the base plate, the bottom of the model. The base plate
was added into the printing folder because it needed prolonged exposure time, rather than
many repeated layers for a shorter time. The base plate can be seen in Figure 15a. This
first layer was exposed the longest to help it stick to the glass slide build plate. This
helps the model stay in position on the glass slide while it was being built.
Given that the cured material must stick to the glass slide throughout the duration
of the printing process, the base plate presents the biggest challenge during the setup of
the print. There is a thin, transparent, oxygen-permeable membrane that is placed on the
bottom of the solution well to help prevent the base plate from curing to the well (Jiang
and Wang 2016). Even with this membrane, the base plate frequently sticks to the
bottom of the solution well instead of moving upwards with the glass slide. The starting
distance between the solution well base and the glass slide build plate is crucial to the
success of the build. The plate needs to be close enough to the base of the solution well
to ensure that enough pressure is applied to the cured liquid so that the plate will stick.
However, if there is too much or too little pressure, the plate will stick to the base of the
solution well and the print would need to be restarted.
38
An example of the different layers of the build can be seen in Figure 15. Figure
15a shows the base plate while Figures 15b-15e show various stages in the structure
build. Figure 15f shows the final layer, which is printed on the top of the model to serve
as a tray to help guide the silicone into the hollow channels of the shell.

Figure 15: Example image layers for 3D printing
a) Base plate. b) Exit chute for excess silicone. b-e) Various layers used to print
final structure. f) Silicone injection plate. g) Image indicates where the layers are
located in the final model.
a) b)
c)
d) e)
f)
g)
e)
d)
c)
f)
b)
39

Fabricating the structure involved two main parts: creating the shell and injecting
the shell with silicone. This method was used to avoid the difficulties associated with
printing soft, elastic material. The printed material, Water2, is hard and inelastic. It was
used to create a shell to inject and cure the softer silicone material. The Water2 solution
used to print the shell is made up of N,N-Dimethylacrylamide (40%wt), Methacrylic
Acid (40% wt), Methacrylic Anhydride (7% wt), Polyvinylpyrrolidone (11% wt), 2,4,6-
Trimethylbenzoyl Phosphine Oxide (2% wt) (Jiang and Wang 2016). The solution is
photocurable and also dissolves in sodium hydroxide. This was ideal for creating softer
structures because it could be used with this method of 3D printing, while also allowing
for easy removal. Exposing certain areas of the solution to UV light hardened the liquid,
causing the print to be built. The solution could then be dissolved once the silicone was
injected and cured to release the positive structure. Before the injection of additional
material, the shells were rinsed with ethanol and placed in a UV cure box for thirty
minutes to ensure that the entire structure was completely cured. The Water2 shells were
then injected with silicone and left to cure for 12 hours at 25ºC. The silicone used was a
tin-catalyzed elastomer (Mold Max NV14) mixed at a ratio of 10:1 silicone to catalyst by
weight. After this, the filled shells were placed in a solution of 1 mol/L sodium hydroxide
and left until the shell was dissolved or easily removable from the actual structure, as can
be seen in Figure 16. Finally, the finished structure was rinsed in deionized water (Jiang
and Wang 2016).
40

Figure 16: Shows the steps of silicone injection
a), c) and e) reflect a rendered version of process of creating the structures.
b), d) and f) give a picture of the actual structure at that stage.
a) and b) show the Water2 shell, unfilled. c) and d) show the structure after the
silicone has been injected into the shell. e) and f) show the final silicone part after
the shell has been dissolved.

a)
b)
c) d)
e) f)
41

Figure 17: Dissolving shells
a) Three samples with dissolving shells in sodium hydroxide.
b) Sample with mostly dissolved shell.

Dissolving the shell in sodium hydroxide presented some difficulty because the
Water2 shell would expand while it was being dissolved, as can be seen in Figure 17a.
The expansion of the shell could stretch the silicone part, which could alter the structure.
a)
b)
42
To avoid this damage, the samples were monitored during soaking. The shells were left
to dissolve and expand for roughly two hours, at which point the Water2 structure was
soft enough to be gently removed from the structure by hand. This helped to avoid
stretching the samples because the shell was expanding more than it was dissolving. It
also sped up the extraction time because the entire shell did not need to dissolve.

2.4 3D Printing Challenges
Although controlled, the process still had many variables. While the CAD models
were all identical, aside from the induced defect, the structures were easily warped during
printing. Since the shape was dictated by the photolithographic image, it was crucial that
the projector was placed at the correct distance and clearly in focus. Day-to-day, the
projector could move or tilt slightly, altering how the image hit the base of the solution
well. This affected the exact dimensions of the structure, since all six of the samples
could not be printed in one day due to time constraints. Even after creating the shell, the
structure was still subject to variability. While rinsing the part with ethanol cleaned out
most of the excess liquid, the final curing process often warped the printed structure,
resulting in one corner being more stretched out than the others. Because not all of the
solution cures completely during printing, it was important to rinse the lattice structure as
soon as it finished printing to avoid having excess liquid cure on, inside, or around the
newly finished print. Additionally, the injected silicone was mixed separately, as it was
needed for each individual shell. Therefore, even though the ratio of catalyst to base in
43
the injected silicone batches was always 10:1, slight variations in each batch could have
contributed to the overall part-to-part variation.
While there may be slight part-to-part differences in the 3D printing method, the
process of 3D printing enables easy fabrication of multiple structures with minimal
amounts of variation. In contrast, tissue samples have significantly more variability even
within the same organism due to the heterogeneity of function within the organ.
Therefore, 3D printed models based on physiological parameters set from compressive
testing can be used to model tissues. By 3D printing various structures, a larger sample
size is much less difficult to obtain. This increases the accuracy of the results given that
more data can be collected, thus decreasing the likelihood of unpredictable error. As a
result, improved access to more precise and applicable mechanical data is available.
Being able to design and print a variety of complex structures with the necessary
amount of detail is promising for further modeling of and comparison to tissue. The
structures are conveniently able to be manipulated and do not take an excessive amount
of time to make. While there are some improvements to be made, this method pushes the
boundaries of what is possible when creating accurate and dependable tissue models. In
the future, additional models of organs will be fabricated to expand this area of
knowledge.

44
Chapter 3: Mechanical Characterization
3.1 Sample Testing
In wanting to compare the mechanical properties of the 3D printed structures to
tissues samples, predominately the pancreas, liver, and kidney, identical testing
conditions were used for ideal comparison. All of these tissue types rely heavily on their
extracellular matrix for their different functions. Therefore, achieving comparable curves
would prove that comparison of the data results is feasible. While the Instron Loadframe
is not ideal for testing biological samples, due to its lack of portability and cleanliness, it
was used in this test because of its convenient setup and manipulability. Compressive
testing of tissue samples requires a disposable component to avoid contamination
between test samples and ensure a clean lab environment, a feat that is not possible with
the metal plates of the Instron and large base. Additionally, a portable setup is necessary
for biological testing given the amount of change the sample undergoes immediately after
resection. Being able to test the samples right away decreases the variability of the test
significantly. Therefore, the pancreatic tissue samples were tested using OFPE because
of its smaller, portable, and disposable components. The sensitivity of optics makes this
a good method for biological testing, and since the optical fiber is relatively cheap, it can
be routinely disposed of after testing to prevent contamination between samples (Hudnut
and Armani 2017). Although the methods for testing the 3D printed samples do not
match those used on the tissue sample, the loading conditions remained the same.
Therefore, a meaningful comparison between the data sets can be made.
45

Figure 18: Picture of Instron compression testing setup with sample.

The samples were tested at 10%, 20%, and 30% strain, providing a varied rate of
compression, 0.067 mm/sec, 0.13 mm/sec, and 0.2 mm/sec, to correspond with the
desired strain, respectively. The samples were tested for a total of 30 seconds, 15
seconds each of loading and unloading. All of the samples were 10 mm x 10 mm x 10
mm so 10% strain corresponds to a 1mm deflection. For 1 mm of deflection in 15
46
seconds, a 0.067 mm/sec rate of compression was used. The same process was used to
determine the rates for 20% and 30% strain. The tests were performed at varying strains
to account for the effect of distinctive rates of compression on different mechanical
properties. 30% strain was used as the maximum strain to avoid permanent damage to
the samples. It is also commonly used to obtain the Young’s modulus in biomaterials,
thus providing a means of comparison to other materials.

3.2 Data Analysis
Looking at the buckling point for multiple strains studied the effect of the rate of
compression on the buckling point; certain materials cannot react as quickly to
compression. Therefore, testing with a high compression rate results in lost information.
Furthermore, some materials appear elastic at 10% strain, while testing them at 30%
strain reflects some viscoelastic properties (Fung 1981).

Figure 19: Sample indexing
a) Base. b) Sample 1. c) Sample 2. d) Sample 3. e) Sample 4. f) Sample 5.
a)
b) c) d) e) f)
47

Figure 20: Compression curves at 10%, 20%, and 30% strain
Graph of the six different structures at 10%, 20% and 30% strain. The difference
in the structures is reflected in the difference in the curves.

The different samples’ response to compression is shown in Figure 20. Across all
the graphs, the base had the highest buckling point, which makes sense given that there
were no induced deformities. This can be seen in the highest of the six peaks, which is
represented by the base color across all strains. Table 1 reflects the highest buckling load
for the base sample while Sample 3 had the lowest buckling load for 20% and 30% strain.
For 10% strain, the lowest buckling load was Sample 5.

Compression at 10% Strain Compression at 20% Strain
Compression at 30% Strain
48

Sample Buckling Point at
10% Strain
Buckling Point at
20% Strain
Buckling Point at
30% Strain
Base 9.3%

0.248N 12.74% 0.261N 12.6% 0.279N
Sample 1 9.1% 0.158N 13.7% 0.168N 12.8% 0.173N
Sample 2 9.3% 0.259N 10.8% 0.254N 10.1% 0.220N
Sample 3 9.2% 0.117N 11.97% 0.115N 14.51% 0.123N
Sample 4 9.4% 0.920N 12.7% 0.202N 14.8% 0.198N
Sample 5 9.2% 0.081N 13.3% 0.120N 14.7% 0.135N

Table 1: Buckling point values for the six samples

Surprisingly, Sample 1, the structure with the least support, did not have the
lowest buckling point. However, this could be due to the movement of the trusses along
the compression plates instead of directly compressing all the trusses of the structure.
Sample 2, which was only missing one truss in a middle position, had the second highest
buckling point. This makes sense, given that this configuration allows the outer trusses to
bare the load equally, thereby creating a stronger structure. The difference between
Sample 4 and Sample 5 was interesting because it indicated that an outside truss was
more crucial to resisting buckling than the middle truss. This is indicated by the higher
buckling load seen in Figure 20 and Table 1.
For each sample, the compression test was run four or five times to prove the
repeatability of the experiment. There was no significant deviation in the data between
runs, which indicates that the sample was not drastically altered between subsequent runs.
This suggests that the compression testing was non-destructive, given the repeatability of
49
the data. Visually, the greatest variance can be seen in the 30% strain graphs, specifically
in Sample 1, in Figure 22. It can be postulated that this was because the structure had the
least support. Missing three trusses, the model was subject to more variation in the
deformation during compression given that there were fewer supporting structures. The
deformation of Sample 1 due to compression can be seen in Figure 21.

Figure 21: Compression of Sample 1 using Instron machine
As the metal plates compress, it is clear from this image that the sample deforms.
The left side of the structure bends outwards and has nothing to hold the trusses in
place given that it is missing three lateral, connecting trusses. In this structure, the
ends of the truss move from their original location in reference to the compression
plate which has the potential to give the sample a higher buckling point than if the
ends of the truss were fixed. This is because a lot of the load goes into the deflection
of the outer trusses which is not directly translated to compression of the sample.

The absence of the trusses left the outside edge unattached to the main body of the
structure. This meant that the defects could bend either toward or away from the lattice
50
structure, meeting similar resistance in either direction. In comparison, the other
structures had more support from additional trusses that prevented this deformity.

Figure 22: Study of Sample 1
a) Graph of 30% strain. Shows variability between runs.
b) Image of Sample 1. Arrows indicate multiple directions outside edge could bend.
Outside edge indicated with blue line.

In the other structures, the deformation of the edge was met with more resistance.
Depending on which direction the edge moved, the connecting trusses either stretched or
compressed. This eventually led to buckling depending on which direction the edge
moved, given that the silicone was elastic enough to allow these changes without
breaking. Having this added resistance helped keep the graphs more uniform between
runs because the added force needed to temporarily deform the additional trusses
prevented the added movement during compression. Interestingly, the structure with the
least variation between runs was not the base model, but rather the two samples with only
one truss removed. As can be seen in Figure 23, the five runs are nearly identical for
both Samples 2 and 3.
a) b)
Compression of Sample 1 at 30% Strain
51

Figure 23: Minimal variance between runs
Compression curves of both samples that are missing only one truss.

Using the same testing conditions as previously sampled tissue allowed the
collected data to be compared to tissue compression tests performed with optical fiber
polarimetric elastography (Hudnut, Babaei et al. 2017). The study looked at the point of
buckling for the samples, rather than the hysteresis, thereby only focusing on the
sample’s response to load before buckling occurred. Since it was unconfirmed that the
samples responded completely elastically, looking at the buckling point of the sample
makes the most sense until the mechanical properties can be better characterized. The
process of buckling looks at the point at which a column, or in this case, a truss, can no
longer bare the compressive force. This can be seen as the highest load sustained by a
structure on a load verses strain graph. This is equivalent to a stress verses strain curve
given that stress equals the force, or load, divided by the surface area, which is not
changing significantly. After the point of buckling, the curve tapers off, which
corresponds to a bend in the column.
Compression of Sample 2 at 10% Strain Compression of Sample 3 at 20% Strain
52

Figure 24: Buckling
a) The figure above shows a beam before (left) and after (right) buckling. The
purple arrow shows the compressive force.
b) Rendering of ideal buckling for base structure. Red indicates most stress, blue
indicates least.

The compression testing results from the Instron were also compared to finite
element analysis, a process that predicts the response of a model to an applied loading
condition to help verify that the achieved results are as expected. The study of the
comparison of the testing data to the ideal models confirmed the accuracy of the testing
methods.

a) b)
53
Chapter 4: Future Directions and Conclusions
4.1 Conclusions
Being able to synthesize comparable models for tissue testing would greatly
improve and standardize research on the mechanics of biological samples. The current
lack of uniformity makes data comparison difficult and fails to account for important
components of tissue extraction and examination. Studying artificial models in parallel
with tissue samples, although initially time consuming due to the comparison of parallel
sets of data, would ultimately lead to greatly improving the ability to produce meaningful
results. Bypassing the need for tissue extraction and testing would increase the
convenience, speed, and accuracy of testing by eliminating some of the issues of
contamination of the samples, as well as the changes in tissue properties that occur after
extraction.
The goal of this research was to create and test biomimetic models using the
process of projection microstereolithography. Even though this method did not allow for
direct synthesis of elastic samples, using this 3D printing method to create hollow shells
proved highly successful. The silicone molds formed from these shells made testing
multiple configurations much more feasible.

4.2 Future Directions
Overall, the 3D printed structures were successfully fabricated and tested under
compression. The attainment of accurate compression testing results indicates a
promising future for comparable samples that utilize the same experimentation methods.
54
Testing the 3D printed samples using optical fiber polarimetric elastography, in addition
to providing a direct comparison to the tissue samples, would also allow assessment of
the different areas of the modeled structure due to the narrower testing area of the OFPE
method. This technique could provide specific focus on the induced defects, in addition
to the comprehensive structure data recorded by the Instron. Figure 25 shows a testing
setup that uses the Instron and the polarimetry system simultaneously. The optical fiber
could rest underneath varying parts of the structure throughout multiple runs, to better
study the effect of missing trusses on the buckling point of various models. While the
initial testing methods provided a sufficient comparison, being able to collect data on
specific locations within each structure would greatly increase the understanding of the
mechanics during compression.

Figure 25: Testing setup using Instron and Optical Fiber Polarimetric Elastography
55
3D printing the parts themselves also proved somewhat challenging. Determining
the initial distance of the glass slide build plate to the base of the solution well dictated
the course of the print and was hard to discern. After starting the print, the adhesion of
the base plate to the build plate also needed to be verified. Often times, the base plate
would adhere to the solution well, thus requiring a print restart. While setup of the print
was not difficult, it took up time that could otherwise have been used for actual
fabrication of samples. Maintaining the transparency of the base of the solution well,
while simultaneously preventing the cured solution from sticking to the base instead of
the glass slide build plate, would drastically improve the 3D printing process.
Standardizing the injected silicone and Water2 solution would also decrease the
variability within the process. By doing this, the inconsistencies in the compression
results could be solely attributed to the induced defects. Creating other base models and
exploring additional defects, such as node removal, would also provide additional
comparable data to increase further understanding.
As with any testing method, more samples would improve the process and
increase the confidence in the results. The study of additional defects to the current base
structure, as well as modeling other base structures could provide more accurate,
inclusive data. Altering the diameter of the trusses would also provide interesting results
by tracking the effect of the diameter to length ratio on the buckling point of the
structure. This could be useful given that the mechanical structure of tissue is comprised
of fibers, which differ in thickness as well as composition (Hudnut and Armani 2017).
Creating a structure that has controlled variability would better mimic these complex
characteristics.
56
This study only focused on the buckling point of the structure, not the hysteresis,
however analysis of what occurs after the buckling point would provide additional
understanding of the mechanical characterization of tissue. Given the complexity of
biological samples, studying simplified models would aid in the comprehension of the
mechanical response of tissues and eliminate some of the inconveniences of tissue testing
such as contamination and time constraints. Additionally, achieving mechanical accuracy
would help progress the field of tissue engineering. This could be applied to the
fabrication of tissue scaffolds, which could either be used for mechanical testing, or be
put toward further stem cell research. These models could potentially be used for testing
various treatments or eventually as replacement organs. Overall the study of the
mechanical properties of tissue through the use of 3D printed models has shown to be
very promising and highly applicable to all aspects of tissue study. The opportunities in
tissue engineering and biomechanics are truly limitless and studies such as these, greatly
contribute to our further understanding of the field.

57
References

1. Ayyildiz, M., et al. (2015). "Effect of normal compression on the shear
modulus of soft tissue in rheological measurements." Journal of the
Mechanical Behavior of Biomedical Materials 49: 235-243.

2. Baker, M. (2016). "1,500 scientists lift the lid on reproducibility." Nature
533(7604): 452-454.

3. Bate-Smith, E. C. and J. R. Bendall (1949). "Factors determining the time
course of rigor mortis." The Journal of Physiology 110(1-2): 47-65.

4. Bonzon, J., et al. (2015). "Rigor mortis at the myocardium investigated by
post-mortem magnetic resonance imaging." Forensic Science International
257: 93-97.

5. Bose, S., et al. (2013). "Bone tissue engineering using 3D printing." Materials
Today 16(12): 496-504.

6. Collins, F. S. and L. A. Tabak (2014). "NIH plans to enhance reproducibility."
Nature 505(7485): 612-613.

7. Cross, S. E., et al. (2007). "Nanomechanical analysis of cells from cancer
patients." Nature Nanotechnology 2(12): 780-783.

8. Fung, Y.-c. (1981). Biomechanics: mechanical properties of living tissues,
Springer Science & Business Media.

9. Guo, T., et al. (2016). "Effect of Dynamic Culture and Periodic Compression on
Human Mesenchymal Stem Cell Proliferation and Chondrogenesis." The
Journal of the Biomedical Engineering Society 44(7): 2103-2113.

10. Hamidi, A., et al. (2017). 3D printing PLA and silicone elastomer structures
with sugar solution support material. 10163: <xocs:firstpage
xmlns:xocs=""/>.

11. Harald, C. O., et al. (2008). "Perfusion-decellularized matrix: using
nature's platform to engineer a bioartificial heart." Nature Medicine
14(2): 213.

58
12. Harrison, M. C. and A. M. Armani (2015). "Portable polarimetric fiber stress
sensor system for visco-elastic and biomimetic material analysis." Applied
Physics Letters 106(19): 191105.

13. Hudnut, A. W. and A. M. Armani (2017). "High-resolution analysis of the
mechanical behavior of tissue." Applied Physics Letters 110(24): 243701.

14. Hudnut, A. W., et al. (2017). "Characterization of the mechanical properties of
resected porcine organ tissue using optical fiber photoelastic polarimetry."
Biomedical Optics Express 8(10): 4663-4670.

15. Jain, E., et al. (2014). "Biomaterials for liver tissue engineering." Hepatology
International 8(2): 185-197.

16. Jiang, Y. and Q. Wang (2016). "Highly-stretchable 3D-architected mechanical
metamaterials." Scientific Reports 6: 34147.

17. Kang, H.-W., et al. (2016). "A 3D bioprinting system to produce human-scale
tissue constructs with structural integrity." Nature Biotechnology 34(3): 312.

18. Kasper, D., et al. (2015). Harrison's Principles of Internal Medicine. New York,
McGraw-Hill Education.

19. Kharaziha, M., et al. (2014). "Tough and flexible CNT –polymeric hybrid
scaffolds for engineering cardiac constructs." Biomaterials 35(26): 7346-
7354.

20. Kurts, C., et al. (2013). "The immune system and kidney disease: basic
concepts and clinical implications." Nat Rev Immunol 13(10): 738-753.

21. Radin, E. L. and I. L. Paul (1971). "Response of Joints to Impact Loading. I. In
Vitro Wear." Arthritis & Rheumatism 14(3): 356-362.

22. Wang, Q., et al. (2016). "Lightweight Mechanical Metamaterials with Tunable
Negative Thermal Expansion." Physical Review Letters 117(17): 175901.

23. Zhang, B., et al. (2016). "Biodegradable scaffold with built-in vasculature for
organ-on-a-chip engineering and direct surgical anastomosis." Nat. Mater.
15(6): 669-+.

Abstract (if available)

Abstract The following thesis outlines the results of a series of studies on the mechanical properties of tissue samples in comparison to three dimensional (3D) printed silicone structures. The mechanical behavior of tissue is hard to study due to the lack of standardization of testing methods. Tissue also undergoes physical changes after extraction, which alter the properties of the samples. Accurate modeling of tissue samples has the potential to provide an easier path to greater understanding of tissue structure. This would lead to more accurate disease modeling through benchtop testing of biomimetic samples, resulting in more enhanced treatments. ❧ The primary 3D printing process utilized during this project is called projection microstereolithography. This is used to create a hollow shell of the simplified, modeled tissue structure. Silicone is then injected into the shell, which, in turn, is dissolved away once the silicone has cured. This leaves the finished structure, which is ready for compression testing. ❧ Testing the compressive properties of tissue yields significant information about the sample. The goal of this project was to be able to mimic these tissue tests with corresponding fabricated samples. This was achieved by compressing six varied structures. The first was the base model, comprised of a patterned XX structure, which was modified to create varied samples with induced defects. From the initial model, other structures were created by removing either one or two trusses in varying locations. The samples were then tested at three different strains, 10%, 20%, and 30%, to look at the effect of different load conditions on the buckling point of the sample. ❧ Being able to accurately reflect the properties of various tissues using synthetic models will help standardize the mechanical testing of biological material and provide an accurate structure for future biological study. Synthetic models would eliminate the time constraints of tissue testing, as well as the sensitivity to contamination. They could also provide cheaper models for testing the effect of various treatments on the body. Furthermore, additional coordination between biological and mechanical technology would aid in the advancement of tissue research.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Development and characterization of hierarchical cellular structures

PDF

Mechanical behavior and deformation of nanotwinned metallic alloys

PDF

Passive rolling and flapping dynamics of a heaving Λ ﬂyer

PDF

Structure and behavior of nano metallic multilayers under thermal and mechanical loading

PDF

Shock wave response of in situ iron-based metallic glass matrix composites

PDF

Novel optimization tools for structured signals recovery: channels estimation and compressible signal recovery

PDF

Viscoelastic characterization of soft biological materials

PDF

Tracking and evolution of compressible turbulent flow structures

PDF

3D inference and registration with application to retinal and facial image analysis

PDF

Development of composite oriented strand board and structures

PDF

An experimental study of shock wave attenuation

PDF

The role of nanotwins and grain boundary plane in the thermal, corrosion, and sensitization behavior of nanometals

PDF

The projection immersed boundary method for compressible flow and its application in fluid-structure interaction simulations of parachute systems

PDF

3D object detection in industrial site point clouds

PDF

Deformable geometry design with controlled mechanical property based on 3D printing

PDF

An analytical and experimental study of evolving 3D deformation fields using vision-based approaches

PDF

Accuracy and feasibility of combustion studies under engine relevant conditions

PDF

3D deep learning for perception and modeling

PDF

Data-driven 3D hair digitization

PDF

Supersonic flow study of a capsule geometry using large-eddy simulation

Asset Metadata

Creator Lash-Rosenberg, Lian (author)

Core Title 3D printing and compression testing of biomimetic structures

School Viterbi School of Engineering

Degree Master of Science

Degree Program Mechanical Engineering

Publication Date 01/19/2018

Defense Date 11/13/2017

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag 3D printing,biomimetic structures,compression,mechanical testing,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Armani, Andrea (committee chair), Bermejo-Moreno, Ivan (committee member), Kassner, Michael (committee member)

Creator Email lashrose@usc.edu,llashrose@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-464033

Unique identifier UC11267137

Identifier etd-LashRosenb-5969.pdf (filename),usctheses-c40-464033 (legacy record id)

Legacy Identifier etd-LashRosenb-5969.pdf

Dmrecord 464033

Document Type Thesis

Rights Lash-Rosenberg, Lian

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA