Summer 2009

Download the 2009 SUNFEST Report.

Project Archive

Hank Bink (Electrical and Computer Engineering) - Lafayette College


Advisor: Cherie Kagan

Abstract: Organic transistor technology holds great promise for creating a conformal, human-safe electronic neural interface. These interfaces must amplify the low, microvolt-range brain signals so they can be utilized in analog and digital applications. Brain signals from sensors must be relayed to the transistor’s gate through the dielectric and semiconductor layers, as well as through an encapsulant which keeps the device shielded in the aqueous brain environment. In order to test their amplification gains, silicon wafer-based ambipolar organic transistors with a pentacene semiconductor were tested under nitrogen in amplifying configurations including common source and cascode. Gains for common source amplifiers with resistors were up to 3.5V/V. Gains for the cascode setup revealed the same results as common source. Both the common source and cascode topologies exhibited very low bandwidth with -3dB points of 35 and 25 hertz, respectively. Parylene C, a biologically safe polymer, is a leading candidate to encapsulate pentacene transistors and serve as a dielectric layer between the devices and sensing electrodes. We tested etching of this parylene as well as the dielectric materials benzocyclobutene (BCB) and spin-on-glass (SoG) using both oxygen and SF 6 plasma etching. Parylene was etched at a rate of 0.2μm/min with O 2 plasma. BCB and SoG did not exhibit useful etching under O 2 or SF 6 plasma. View Paper | View Slides

Allison Connolly (Biomedical Engineering) - Johns Hopkins University


Advisor: Rahul Mangharam

Currently, there is no formal method for the development and testing of medical device software, such as that used in pacemakers and implantable cardioverter-defibrillators (ICD). A large majority of device recalls are due to failures in the software that went undiscovered during product testing. For example, safety recalls of pacemakers and implantable cardioverter defibrillators due to firmware (i.e. software) problems between 1990 and 2000 affected over 200,000 devices, comprising 41% of the devices recalled. In order to preempt these failures, the device companies and the regulatory agencies, such as the FDA, need a better way to formally and functionally verify these devices before bringing them to the market. The heart model outlined in this paper is a tool used to simulate, test and validate these devices across multiple modalities in a plug-and-play manner. By synthesizing a large number and variety of intra-cardiac electrogram and derived external electrocardiogram signals, the model will create a database well beyond the scope of the MIT-BIH ECG database, the current standard for most cardiac medical device algorithm testing. This heart model allows for more extensive formal and functional testing of pre-market cardiac medical devices to detect flaws before the devices are implanted in patients. View Paper | View Slides

Phillip Dupree (Mechanical Engineering) - Columbia University


Advisors: Daniel E. Koditschek, Galen Clark Haynes

An important step in the autonomization of robots specifically designed for mobility is autonomous navigation: the ability to navigate from a current position to a programmed point, without the manual control of a human user. The object of this summer research was the autonomization of the robot RHex, a highly mobile hexapodal robot built in the GRASP lab of the University of Pennsylvania, which was accomplished by the integration of a global positioning system (GPS) module into the robot. The GPS module gave the robot the ability to follow a “breadcrumb” path of GPS way-points. Once the GPS data was parsed, the coordinates of both the robot’s location and the path of waypoints were converted into flat-earth approximate Cartesian coordinates, and then inputted into a linear control system. Once this was accomplished, the robot had the ability to “know” its current position and navigate from it to any programmed point, providing there were no obstacles in its path. View Paper | View Slides

Katherine Gerasimowicz (Bioengineering) - University of Pennsylvania


Advisors: Dr. Jay N. Zemel, Dr. Babette Zemel

ABSTRACT: Developing strong bones early in life reduces the risk of osteoporosis in the future. Various types of physical activity have been reported to produce osteogenic effects in children. However, current tools used in bone development research are unable to provide convenient and accurate measurements of the loads experienced in long bones throughout a child’s regular daily physical activity. We have devised an inconspicuous system that can be embedded in a child’s shoe to monitor and store force measurements during the course of a child’s normal wakened activity. This in-shoe physical activity dynamometer, Foot-PAD, has been in development since the summer of 2004. The last model prior to the current research consisted of a circuit that amplified and converted electrical signals from polyvinylidene fluoride (PVDF) piezoelectric film sensors into digital force measurements. PVDF sensors are most sensitive to horizontal forces along the surface of the foot rather than forces directly transmitted to the foot. Repeated efforts to convert the normal force to a horizontal force were unsuccessful in the past. The Emfit Ltd. piezoelectret sensor has been developed with similar charge displacement properties but with the ability to measure vertical forces. The primary accomplishment of this development phase, therefore, was the incorporation of piezoelectret sensors into the system and appropriate modification of the circuit design. Tests with a custom-made mechanical testing device and squat jumps confirmed that the piezoelectret sensor could accurately measure vertical forces. View Paper | View Slides

Willie W Gonzalez (EE) - University of Puerto Rico, Mayagüez


Advisor: Dr. Daniel D. Lee

A key characteristic in any autonomous system is stability. Without stability robots could not work properly. To achieve stability first the robot needs the means to know its position with respect with its original so that a correction in position can be obtained if necessary. This is true for the type of INS (inertial navigation system) known as dead reckoning. In this INS the current position of the robot or system is calculated from measurements of the acceleration and knowledge of its original position.

Some IMUs (inertial navigational units) consist of a couple of accelerometers and gyroscopes. With this IMU acceleration can be obtained. On Strap-down navigational system, this IMU are directly attach to the system so that their measurements correlates with those of the system, so that they can be used to calculate the position of the system. Throughout this document a method on how to obtain the necessary measurements from the IMU and how to apply them, will be discuses. View Paper

Sarah Koehler ( Engineering Science) - Cornell University


Advisors: Dr. Mark Yim and Dr. Daniel E. Koditschek

Manually tuning a robot’s gait so that it can walk quickly and efficiently is a time-consuming and tedious task. For this project, CKBot (a modular robot system) has been configured into a centipede-inspired configuration with six legs. The project started with a few manually tuned gaits which follow an alternating tripod pattern. The purpose of this project is to automate tuning of the robot’s gait such that it optimizes a factor such as specific resistance, speed, or power. Most of the effort in this project has gone into setting up the framework for such optimization trials. Each optimization trial runs the centipede back and forth and adjusts the six parameters that change the robot’s gait by using the Nelder Mead optimization method. The end result after optimization is a gait with minimal specific resistance, maximum speed, or some other optimal factor. View Slides

Linda McLaughlin (Electrical Engineering) - Community College of Philadelphia


Advisors: Jay Zemel, Jorge Santiago, David Graves, Michael Mauk

ABSTRACT: Photovoltaic cells and biofuel technologies have never been fused together to the best of our knowledge. The long term goal of this project is to use these two technologies to develop a prototype to convert the sun’s radiation to different forms of power. The purpose of this present study is to design a PBR that provides light and dark periods the algae need in their photosynthetic process to increase algal growth. For this reason, light shielding (dark regions) was supplied by separated solar cells placed periodically over the PBR. Flat PBRs were constructed. A control PBR without photovoltaic cells provid ing dark periods was set up and exposed to sun radiation. Data obtained was compared with the one that had solar cells upon it. Cell concentration decreased in the system without solar cells (control) on a sunny day. Preliminary observations suggest that the Light/dark period PBR system led to an increase of cell growth under the same conditions. Yet, because of minimal data acquisition and presence of limiting factors such as CO2 depletion, feedstock supply, degassing system, and temperature control, future investigation will be needed. Further studies, in which photovoltaic cells will be used with this dual purpose: converting light into electrical energy and provide the light shielding the algae need to improve growth, will need to be done.

Ieva Narkeviciute (Chemical Engineering) - University of Massachusetts Amherst


Advisor: Jay Zemel, Jorge Santiago, David Graves, Michael Mauk

ABSTRACT: Photovoltaic (PV) cells and photobioreactors that grow algae for the production of biodiesel have never been combined in a hybrid system even though they both use the sun as a source of energy. The goal of this study was to design a photobioreactor that combines modular PV panels for dark and light (DL) periods to increase algal photosynthesis. Simultaneously, the PV panels produce electrical energy that will power the system and any excess power could be fed to the electrical grid. Two bioreactors were constructed. One reactor was not covered by solar cells and served as a control for the experiment. The second reactor had 90% of its surface covered by solar cells to provide dark periods. One goal was to reach comparable or higher yields for the DL reactor in comparison to the uncovered reactor. Another was to examine whether short, cyclic DL periods increase growth. Data acquisition was minimal, but preliminary results show that the system without DL periods had a longer doubling time than the one with the DL periods, proving that the cycles increase growth.

Jeffrey Perreira (Mechanical Engineering) - Lehigh University


Advisor: Dr. Robert Mauck, Graduate Student: Megan Farrell

ABSTRACT: Mesenchymal stem cells (MSCs) harvested from bone marrow tri-differentiation potential into osteoblasts, adipocytes, and chondrocytes along with portraying a variety of phenotypes. MSCs are a promising cell source for cartilage tissue engineering, but MSC seeded constructs have yet to match the mechanical properties of chondrocyte seeded constructs. The basis of this study is to use imaging techniques on MSCs that have already been developed to analyze chondrocytes and native articular cartilage to provide information on matrix production, cell response to load, and cell mechanical properties. This paper displays two studies. The first study used micromechanical analysis with florescent microscopy on MSCs and cartilage to study the local and bulk mechanical properties and development of the cells’ extra cellular matrix (ECM). The hypothesis was that comparing the local mechanical properties to the bulk properties of MSCs and chondrocytes will explain why the MSC constructs are weaker than the chondrocyte constructs. The second part of the study used confocal microscopy to analyze cell deformation as a function of matrix production over time. The cell deformations were analyzed via a customized Matlab (Mathworks) program to act as a standardized analysis method. Analysis when complete of cell, local, and bulk tissue mechanics is expected to provide insight into the subpar bulk mechanics found in MSC constructs and determine the heterogeneity of local matrix properties. View Paper | View Slides

Andrew Townley (Electrical Engineering) - University of Pennsylvania


Advisor: Gianluca Piazza

ABSTRACT: In recent years, energy harvesting using piezoelectric materials has become a very popular research topic. Various device sizes and structures have been tested, but it is difficult to compare power measurements as device fabrication and experimental methods vary from paper to paper. In an effort to standardize comparisons in spite of these changing parameters, the dependence of generator power output on device dimensions has been investigated.

Though MEMS scale devices have been produced, comparatively little work has been done using aluminum nitride (AlN). This project utilizes AlN due to its ease in processing and potential for on-chip integration. By operating at a MEMS scale, the benefit is that arrays of piezo generators can be placed on the same die. With the process advantages of AlN, a long term goal of an integrated power-harvesting chip becomes feasible.

However, theoretical results of scaling predict that raw power output and even power per unit volume will decrease with scaling. This indicates that a single large generator, taking up the same area as several small generators, would produce a noticeably larger power output.
Due to time constraints, no new generators could be fabricated within the time span of the project. An existing piezoelectric cantilever was used to verify the theoretical predictions of resonant frequency and static deflections under applied voltage. These predictions agreed quite closely with the observed results. However, no measurable electrical response could be found while exciting the beam with an electromagnetic shaker device. A similar experiment was performed using an AFM to directly excite the beam, but again the electrical response was difficult to characterize.

While the results of the experiments were not optimal, the difficulty in measuring the electrical response of the beam demonstrates the design challenges involved with energy harvesting on a small scale. Piezoelectric generators rely on resonance to generate useful quantities of power, and power output is highly sensitive to the frequency of the physical vibrations applied. While generators of this type could be useful if targeted to a specific application if the frequency of environmental vibrations is known, a more versatile approach would use a different design to reduce the frequency sensitivity. Broad-band designs, using either non-resonant or self-tuning structures, would be able to harvest energy much more efficiently in changing environments. View Paper

Desirée E. Velázquez Rios (Mathematics) – University of Puerto Rico at Humacao

MIMOSA: Mine Interior Model Of Smoke and Action

Advisors: Norman I. Badler and Jinsheng Kang

ABSTRACT: The team goal for the summer is the simulation of underground coal mine fires, and the study of the physiological reactions occurring during evacuations. I undertook the task of searching for new ways to modify the map creation tools used by the original crowd simulation software using the Crowds with Aleatoric, Reactive, Opportunistic, and Scheduled Action (CAROSA) framework. I also modeled the mining equipment used to simulate a more realistic coal mine situation. First, a literature search on the web was done for software called Level Editors and Game Engines. Then, modeling the mining equipment, rigging the miner mesh model with a skeleton and animating both equipment and miner was done. These tasks used two pieces of Autodesk software; Maya and Motion Builder. Additionally, Maya was used to create the cell-portal mine map that the virtual miners use to move about in the space. Once the models and animations were created, they would be used for the simulation. With the simulation results we can validate a study made by U.S. Department of Health and compare the physiological models visualized in the simulation with the results of the study.

Valerie, Walters (ESM) – Virginia Polytechnic Institute and State University


Advisor: Dawn M. Elliot, PhD

ABSTRACT: Many people suffer from lower back pain, which can be caused by disc degeneration. We are studying the properties of non-degenerated and degenerated human lumbar intervertebral discs. The overall objective of this study is to quantify strains in the disc when it is under torsional and compressive loading, simultaneously. This will be done by using magnetic resonance images of discs before and after loading and advanced normalization tools, which is an image normalization technology. For this summer, the objective is to design a device that will load the specimen as previously described and be compatible with the magnetic resonance imaging machine. The protocol will consist of incrementally torquing the specimen while in constant compression. The torque and the compressive force will be monitored via a load cell throughout the test. We expect displacement results to agree with previous studies that have been done in standard mechanical testing equipment. We also hypothesize that the outer annulus fibrous will be the stiffest part of the disc, and that the posterior region will be stiffer than the anterior region. The strain results that we expect to obtain do not have a previous test to compare to; this is the first study that will quantify internal disc strains while the disc is in torsion. We expect the results to provide useful information about the properties of the human intervertebral disc. View Paper | View Slides