Fall 2005 – Spring 2006

HSGC Report Number 06-14

Compiled in 2006 by HAWAI’I SPACE GRANT CONSORTIUM

The Hawai’i Space Grant Consortium is one of the fifty-two National Space Grant Colleges supported by the National Aeronautics and Space Administration (NASA). Material in this volume may be copied for library, abstract service, education, or personal research; however, republication of any paper or portion thereof requires the written permission of the authors as well as appropriate acknowledgment of this publication.

This report may be cited as Hawai’i Space Grant Consortium (2006) Undergraduate Fellowship Reports. HSGC Report No. 05-13. Hawai’i Space Grant, Honolulu.

Individual articles may be cited as

Author, A.B. (2006) Title of article. Undergraduate Fellowship Reports, pp. xx-xx. Hawai’i Space Grant Consortium, Honolulu.

This report is distributed by:

Hawai’i Space Grant Consortium Hawai’i Institute of Geophysics and Planetology

University of Hawai’i at Manoa 1680 East West Road, POST 501

Honolulu, HI 96822

TABLE OF CONTENTS

Page Forward………………………………………………………………………………….… i POWER GENERATION AND DISTRIBUTION SYSTEM DESIGN FOR THE LEONIDAS CUBESAT NETWORK………………………………………………….….1 Justin M. Akagi University of Hawai‘i at Mānoa ARCHAEOLOGY ON EASTER ISLAND: REMOTE SENSING FOR LOW-IMPACT ARCHAEOLOGICAL STUDY OF RAPA NUI'S (EASTER ISLAND) MONUMENTAL STONE ARCHITECTURE …………………………...…...….....…... 7 Matthew J. Bell University of Hawai‘i at Mānoa PLANNING OF KEPLERIAN ORBITS: APPLICATION TO PERIODIC TRAJECTORIES ……………………………………………….….….….….….….….. 13 Vann Michael Bennett University of Hawai‘i at Mānoa RECONSTRUCTION OF THE EVOLUTION OF GALAXIES USING THE EXTENDED-PRESS-SCHECHTER MODEL WITH GLOBULAR CLUSTERS………………………………………………………………………………19 Eli Bressert University of Hawai‘i at Hilo FIRST INFRARED SPECTROSCOPIC CHARACTERIZATION OF THE Ge2H(3,5) MOLECULES AND Ge2D(3,5) RADICALS IN LOW TEMPERATURE GERMANE MATRICES………………………………………………………………………………26 William Carrier University of Hawai‘i at Manoa MECHANISTIC STUDIES OF TRIFLUOROMETHYL SULFUR PENTAFLUORIDE SF5CF3: A GREENHOUSE GASFIRST INFRARED SPECTROSCOPIC ………..…...33 William Carrier University of Hawai‘i at Mānoa DECADAL VARIABILITY IN SLOPE STREAK ACTIVITY ON MARS ………..…. 39 David Gremminger University of Hawai‘i at Mānoa SYSTEMS INTEGRATION AND STABILIZATION OF A CUBESAT ….…………..44 Tyson Kikugawa University of Hawai‘i at Mānoa

IMPROVING THE DECONVOLUTION METHOD FOR ASTEROID IMAGES: OBSERVING 511 DAVIDA, 52 EUROPA, AND 12 VICTORIA ….……………...…..52 Z Robert Knight University of Hawai‘i at Hilo DETERMINING THE DISTRIBUTION OF YOUNG BINARIES IN THE ORION NEBULA CLUSTER……………………………………………………. ….…………..59 Sarah Knights University of Hawai‘i at Hilo A THEORETICAL INVESTIGATION ON THE KINETIC AND RADIATIVE EXTINCTION OF SPHERICAL DIFFUSION FLAMES IN MICROGRAVITY…….. 63 Kin Wai Leung University of Hawai‘i at Mānoa INTELLIGENT SENSOR NETWORK FOR EXTREME ENVIRONMENTS….…….. 71 Mary Liang University of Hawai‘i at Mānoa HARDWARE OPTIMIZATION OF THE TRIGGER UNIT FOR RADIO FREQUENCY FOR ULTRA HIGH ENERGY NEUTRINO DETECTION IN ANTARCTICA…...…..77 Brandon A. Merz University of Hawai‘i at Mānoa EFFECTS OF LOW REYNOLDS NUMBERS ON THE AERODYNAMICS OF MICRO-AIR VEHICLES….………………………………………………………...….. 83 Shelly A. Migita University of Hawai‘i at Mānoa LAVA FLOWS IN THE THARSIS REGION OF MARS: ESTIMATES OF FLOW SPEEDS AND VOLUME FLUXES………………………………………....…………..88 Carolyn Parcheta University of Hawai‘i at Mānoa MAPPING THE PREHISTORIC STATUE ROADS ON RAPA NUI USING REMOTE SENSING SATELLITE IMAGERY ….……………………………………………..…. 93 Gabe Wofford University of Hawai‘i at Mānoa INTELLIGENT SENSOR NETWORKS FOR EXTREME ENVIRONMENTS……..... 98 Faye S.Y. Yuen University of Hawai‘i at Mānoa LEONIDAS SATELLITE CONCEPT STUDY REPORT ………………………….....104 LEONIDAS Team University of Hawai‘i at Mānoa

Forward

This volume contains eighteen reports from Hawai’i Space Grant Undergraduate Fellows at the University of Hawai’i at Manoa and the University of Hawaii at Hilo. The students worked on their projects in the Fall 2005 and Spring 2006 semesters under the guidance of their faculty mentors. We congratulate all of the students for their outstanding reports and warmly thank their faculty mentors for generously supporting the Fellowship Program.

The Hawai’i Space Grant Consortium is supported by NASA through its National Space Grant College and Fellowship Program with matching funds from the University of Hawai’i. The goal of the program is to strengthen the national capabilities in space-related mathematics, science, and engineering and to prepare the next generation of space scientists. All of the students’ projects are related to the goals of NASA’s Strategic Plan.

For more information about the Fellowship Program, please visit our website: http://www.spacegrant.hawaii.edu/fellowships.html Edward R.D. Scott Associate Director, Fellowships

1

POWER GENERATION AND DISTRIBUTION SYSTEM DESIGN FOR THE LEONIDAS CUBESAT NETWORK

Justin M. Akagi

Department of Electrical Engineering University of Hawai`i at Manoa

Honolulu, HI 96822

ABSTRACT

The Power Generation and Distribution (PGD) system design of the University of Hawaii Small Satellite Program’s Phase III, Ho`okele (Way Finder), is described. The design progression of the PGD system is presented to provide a working reference for small satellite developers. The system is described using the fundamental blocks: Power Generation, Energy Storage and Power Management. Power generation and system-level power management are discussed from both hardware and software standpoints. The final design of the PGD system for Phase III is also presented.

INTRODUCTION

The University of Hawaii’s Small Satellite Program was created with the intent of

promoting high-technology research with the State of Hawaii. By creating a university-level satellite program, undergraduate and graduate students will have the opportunity to work on real-world engineering projects. One of the major advantages of these student-driven satellite projects is the cost – small satellites may be developed for a fraction of the cost required to develop larger, conventional satellite systems. In addition, small satellites are designed to support experimental payloads, and may be completed within a much shorter period of time. The ability to support experimental payloads also offers an excellent opportunity for student developers to work collaboratively with the scientific community and high-tech industry.

TECHNICAL OVERVIEW

The Power Generation and Distribution System is responsible for generating and

supplying power for the satellite bus and payload. Regardless of the specific design requirements, the basic building blocks for any small-satellite power system are essentially the same. Energy received from the solar cells is used to power the system during sun-on periods, and to recharge the battery pack for sun-off periods. During solar eclipse, the battery is used as the primary power source. The main power line (connected to the solar cells and battery) feeds into a number of DC-DC power converters, which provides the necessary supply voltages for the satellite’s electronics. A battery monitor is also integrated into this system to measure sustainable power consumption and facilitate in power management. Fig. 1 shows a basic block diagram of the power system.

2

Figure 2: 2D Solar Cell Layout. Solar cells labeled as SCxx. Empty blocks represent space occupied by other

system hardware.

Figure 1: Basic block diagram of PGD system. The basic building blocks within this block diagram are present in

any autonomous small-satellite power system.

Although every autonomous satellite system has these basic building blocks, each system requires unique power components and circuitry. Specific component selection and circuit design is dependent on a number of system/payload requirements and hardware specifications.

POWER GENERATION

Spectrolab improved triple junction solar cells were selected based on their high-

efficiency (26.8%) characteristics [1]. The solar cells are connected as pairs in series, with all of the pairs connected in parallel. This layout is necessary because of the dimensions of the small satellite (10x10x15 cm). Each of the square faces of the satellite (10x10 cm) could fit two solar cells, and each of the elongated faces (10x15 cm) could fit four cells. (The two solar cells in each series pair would be located on the same side.) This design ensures that regardless of the satellite’s orientation, both solar cells in every series pair would receive the same amount of solar energy.

However, after specifying this solar cell layout, it was determined that external hardware would require a considerable amount of space on the two square faces, and no solar cells could be placed on either of these two faces. Fig. 2 shows the finalized solar cell layout.

The solar cell circuit

configuration is shown in Figure 3. Pairs of solar cells (on the same face) are connected in series, and all of the pairs

3

are connected in parallel.

Figure 3: Solar Cell Circuit Connections. Bypass diodes are built into the solar cells to protect from series connection power drain. Blocking diodes are integrated to prevent power drain caused by unpowered pairs.

ENERGY STORAGE

Small-satellite developers must also consider the issue of energy storage. During periods of orbit where the satellite is in view of the sun, the power system uses the readily available solar energy. However, during periods of solar eclipse, the satellite must use previously stored energy. To maximize the system’s energy storage capacity, lithium-ion batteries are used for our satellite system. Fig. 4 shows a visual comparison of the energy densities of lithium-ion (Li-ion), nickel-metal hydride (NiMH), and nickel cadmium (Ni-Cd) batteries. As the figure shows, Li-ion batteries have a greater energy density (both gravimetric and volumetric) than Ni-Cd and NiMH batteries.

Figure 4: Comparison of Energy Density vs. Battery Type (Li-Ion, Ni-Cad, Ni-MH) [2]

Based on refined battery characteristic requirements, the UR18650F lithium-ion battery

was specified for energy storage for the power system. Based on power requirements, a 2-cell battery pack (4200mAh) was selected. Some of the parameters that affect battery selection include: operating temperature range, maximum load current, charge/discharge cycles, time-averaged depth of discharge, and power management schemes.

4

POWER MANAGEMENT

The power requirements for Ho`okele have been modified at least four times due to changes in the payload system design. Although the experimental payload is not a part of the main satellite bus, the power system must be designed to satisfy the specified payload power requirements. Payload system redesign and corresponding power requirement modifications led to many intermediate design changes. Due to the numerous design changes, the high-level design of the PGD system was made modular, so that the power system circuit could be easily modified to accommodate any power requirement changes. These modules may also prove useful for our subsequent satellite projects, in which they may be easily integrated into the power systems (based on design requirements). Table 1 shows a list of the on-board electronic components with corresponding supply voltage and load current requirements.

Table 1: Satellite Power Requirements

The power converter design for the PGD system is shown in Fig. 5. The converter

circuits are designed to satisfy the power requirements specified in Table 1. A 3.3V, 600 mA regulated source [9] powers the GPS unit (Payload), and the I2C to 1-wire bridge (PGD). A 5.0V, 1.2 A source powers the satellite’s primary microprocessor board, a Linux-based embedded system, and the Ethernet camera (Payload). A 5.4V, 1.2A source supplies the transceiver (TTC) with the required power, and a 6.5V, 300mA source powers the nIMU (Payload).

5

Figure 5: Power Converter Design Layout

After designing the power converter layout, a management scheme needed to be created

to manage the limited on-board power supply. Unlike many earth-bound systems, there are no power outlets that the satellite can plug into to satisfy its power needs. Instead, mechanical and electronic switches are required to manage the power consumption of the on-board systems. In addition to the power-switching network, a battery monitor is necessary to facilitate in battery management during normal operation of the satellite. Fig 6 shows the complete power system design, with power management devices.

Figure 6: Complete PGD System Design with Integrated Power Management Devices

6

During the satellite’s normal operation, on-board electronic components (camera, transceiver and nIMU) are powered on and off to avoid unnecessarily high power consumption. This power-switching design is necessary to ensure that sufficient power levels are maintained throughout solar eclipse periods. The power switches are implemented in hardware using Fairchild Semiconductor P-channel MOSFETs. The MOSFET is capable of regulating current from the source (input) to the drain (output) using a logic level control signal (gate). When the control signal is set high (5 V), the power switch cuts off power to the electronic component; when it is pulled low (0 V), the current may flow from source to drain. Power management decisions for normal operation flow are based on the current battery status, and will be performed by the satellite’s primary microprocessor. The DS2780 battery monitor is capable of performing battery voltage, current flow, temperature, and battery capacity measurements. The telemetry data from the battery monitor is used to facilitate satellite operation flow decisions (e.g.: if there is enough power to operate electronic devices or perform user-defined missions).

CONCLUSION

The third phase of our Small Satellite Program is reaching its completion goal. In the last few months, our undergraduate team has done a good job designing all of the satellite subsystems. Each of the groups, though nominally different, needed to work together to integrate the entire system. The Power Generation and Distribution System has been a huge task (individually, and for system integration), but our group has done good work designing the system to ensure that it satisfies all of the satellite requirements.

ACKNOWLEDGEMENTS

I have been working on University of Hawai`i Small Satellite projects for four years,

starting with Phase I of the program. The practical experience provided by this satellite research has been an excellent opportunity for me. I would like to thank NASA, the University of Hawai`i Space Grant Consortium, and all the other sponsors of our Small Satellite Program. I would also like to thank Dr. Wayne Shiroma for working so hard to establish the Small Satellite Program at the University of Hawai`i, and for providing his technical experience and guidance throughout the lifetime of the program.

REFERENCES

[1] Spectrolab webpage. http://www.spectrolab.com. [2] Sanyo Energy (USA) Corporation webpage. http://www.sanyo.com/batteries. [3] ACME Systems webpage. http://www.acmesystems.it. [4] Maxim/Dallas webpage. http://www.maxim-ic.com. [5] Microhard Systems Inc. webpage. http://microhardcorp.com. [6] MEMSense webpage. http://www.memsense.com. [7] Garmin International Inc. webpage. http://garmin.com/index.jsp. [8] Axis Communications webpage. http://www.axis.com. [9] Linear Technology webpage. http://www.linear.com/index.jsp.

7

ARCHAEOLOGY ON EASTER ISLAND: REMOTE SENSING FOR LOW-IMPACT ARCHAEOLOGICAL STUDY OF

RAPA NUI'S (EASTER ISLAND) MONUMENTAL STONE ARCHITECTURE

Matthew J. Bell Department of Anthropology

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

This study uses remote sensing techniques to study ahu (monumental ceremonial

platforms) on the island of Rapa Nui (Easter Island). The first phase of the project used commercial satellite images to locate ahu and create a comprehensive GIS database to be corroborated during future ground survey. To supplement the satellite images, a second phase of the study evaluated kite photography - a low-cost, non-destructive, low-impact and culturally sensitive way to supplement the resolution of the satellite images. Kite photography was used to photograph ahu and ground targets were used to mosaic photographs together. From the resulting image, traditional archaeological plan view maps were made. The micro and macro survey of archaeological features in this study will contribute first towards protecting the famous heritage sites on Rapa Nui for the Rapanui people and for future research. Data will be used in continuing studies that aim to scientifically understand cultural and environmental change on the island.

INTRODUCTION

Rapa Nui is an oft-mentioned place, but what is not widely appreciated is how little basic

knowledge actually exists about the island. Though archaeologists have worked there for decades, there are still basic holes in simple locational data. Traditional large area survey requires long hours with intensive human labor, and so the cost of filling those holes remains high. Additionally, many structures, such as ahu, are in very poor structural condition. Not only would a detailed survey of an individual structure dramatically expand a survey’s time frame, it would also endanger the subject of the study itself. Ahu, in particular, are complicated because of their nature – they are traditional familial burial platforms upon which no one is allowed to step. This study attempts to address these problems in two phases – a macro study of the ahu location via satellite image and a micro study of a few sample ahu with kite photography to document their condition, construction and to provide a vantage point from which to draw traditional archaeological maps. With this project as a precedent, it is hoped that a good, publicly available digital database of ahu locations will be refined over time. Additionally, if more ahu and other large structures are studied with kite photography (or other aerial means), hopefully the impact on the archaeology itself and the cultures to which archaeologists are responsible will be reduced.

8

METHODS

For the satellite image survey a total of three commercially available QuickBird images of the island were loaded into ArcGIS. East-west survey lines were created in the computer at intervals of 250 meters and the region between these lines was observed systematically from south to north across the entire island. Potential ahu locations were recorded in UTM coordinates in a different file. Potential ahu were identified based on specific criteria such as size, shape and tone. Additionally related structures, coastal orientation and locations known to be ahu by the author or third party publications were used to help verify that observed phenomena were indeed ahu (Martinsson-Wallin 1994, Cristino, et al. 1981).

During the second phase a reliable method for taking good quality photographs from a kite was explored. A riveted aluminum rig constructed for kite photography by a previous student was modified and field-tested to suggest the best design for a new rig for this project. Design goals for the new rig included durability, easy operation, flexibility in application, and consistent photographic results. Durability was the primary consideration in material selection, as the rig is intended to be backpacked into remote locations and used by largely untrained crews of field school students or Rapanui assistants. Ease of operation was also desirable to decrease the learning curve of using the equipment and to speed deployment by experienced operators. The rig was designed to be adjustable to accept different cameras in the event of failure of the primary camera. Finally, consistent results were a mandatory feature of the rig. A four-point “Picavet” suspension system similar to those used by hobby kite photographers was used to minimize twisting and swaying and to self level the rig via pulleys (Haefner 2005).

The resultant rig is a polycarbonate shell (1) enclosing an aluminum and fiberglass frame (2) that holds the radio receiver (3), shutter activating servo (4), radio power supply (5) and a

mounting plate for the camera (6). All joints in the rig are either bonded with epoxy resin or are held together with nylon zip ties. The latter are much lighter than other fasteners, allow for flexibility in certain joints and are easy to replace in the field. The rig also has a plastic laminated aluminized Mylar shield that closes the hole in the top of the polycarbonate shell. It provides protection for the camera and radio receiver against the sun and unpredictable light rain showers, but allows easy access to the camera to adjust settings.

FIGURE 1 – Camera Rig

A diagram of the camera rig as described in the text. The pulleys, connected with small line to the rig, create a four-point suspension.

Pulleys

1

2

5

3

4

6

9

All aerial photographs are subject to several types of distortion, which reduces their usefulness in making accurate measurements. The first type is optical distortion from the camera itself. Most commercial aerial photographs are taken with very large lenses and so optical distortion is minimal. For this application, a light, compact camera was more ideal. A lighter camera allows for a larger range of lift producing wind. Additionally, smaller consumer digital cameras are considerably cheaper, critical for a limited budget project. As a result, the smaller lens incurs more optical distortion. Controlled test photographs found that warping of the images occurred largely at the edges of the photograph. Also, tests with a freely available high quality, open source program called Panorama Tools showed that software correction of the optical distortion produced good results, with only minor residual distortion. This software was used to process all photographs before use (Dersch 2005).

Another type of distortion to which all aerial photographs is subject is tilt in the camera. This problem was primarily addressed in the four-point rig suspension and self-leveling system. Secondarily, a survey grade GPS unit was used to take UTM coordinates of cloth targets pinned to the ground in the target area. Transforming the photographs to fit the GPS coordinates created an image more true to real life. Any photographs that required heavy transformation to fit the coordinates were assumed unacceptably tilted and discarded.

The final type of distortion is that of radial displacement of objects with height outward from the center of the photograph. This displacement increases with distance from the center of the photograph. For a continuous surface such as a beach, the GPS ground targets allow for reasonable transformation when there are ample targets in a photograph. For discrete objects such as palm trees, radial displacement was problematic. As much as possible, photographs were cropped to fifty percent of their original width and sixty-seven percent of their original height to remove the areas with the heaviest radial displacement. However, because objects like palm trees are not the subject of study, considerably higher error in alignment was tolerated than for archaeological structures. Given a good method for recording the altitude of the camera at the time each photograph is taken and a good digital elevation model of a target location, this most elusive form of distortion can be largely eliminated with the creation of an orthophotograph using specialized software (Wingert 2005). This, however, was outside of the scope of this project, though it would be advisable to explore this technique in a future project.

Photographs that showed good exposure, composition and low distortion were used in the final stage of this project. These photos were imported into ArcGIS and georeferenced using the GPS coordinates. They were layered into a mosaic to give a continuous image of an area under study.

RESULTS

The satellite survey resulted in the identification of 331 ahu locations. Figure 2 shows the distribution of these locations around the island. These locations are, of course, based only on generalized identification criteria. Some of these points may turn out to be modern structures and other ahu, especially small ones, may remain hidden. This survey is, however, an important first stage to creating a comprehensive database. As has been done with previous satellite image surveys of roads and agricultural features on Rapa Nui by Dr. Terry Hunt, Dr. Carl Lipo and their respective students, this ahu survey will be subjected to ground-truthing during upcoming

10

field seasons (Lipo and Hunt 2005; Bradford, et al. 2005). At a minimum, a random sample of the identified points will be located to estimate the success rate of identification. However, because ahu are large structures and are prominent in ancient and modern Rapanui culture, over time it should be possible to verify all ahu points with the help of hand-held GPS units either as a stand alone survey, or in conjunction with other survey.

During the summer of 2005, a research trip to Rapa Nui was taken to photograph three different locations of archaeological interest. The most extensive photography was undertaken at Anakena beach on the north shore of the island. The resulting photo mosaic can be seen in Figure 3 on the following page. Aerial photography was also conducted at an ahu on the north coast and one on the south coast. These mosaics were considerably smaller, consisting of only a few photographs.

For ahu of particular interest, close-ups of the photo mosaic were traced to produce a planimetric map of the structure in a traditional archaeological style in a simple representation of its construction and condition. Large boulders and cut stone down to cobbles approximately 20 cm in diameter were depicted separately. Smaller stones were not distinguishable from each other at the resolution provided by the kite photographs. Instead an area showing a pattern suggesting scatterings of small stones was represented with a generic pattern on the map, as is common in archaeology.

In some cases, there were limitations to what data the aerial photographs could provide. For example, whether large stones were uprights or not was not always clear. Also, type of stone was sometimes not apparent. This caused some problems in identifying which stones were structural, which were fallen remnants of moai (statues often topping an ahu), or pukao (stones set on the heads of the moai), all of which are composed of different local stones. However, were greater detail desired in the maps, a quick ground reconnaissance would easily reveal such details. As with all remote surveys, the final product could be improved with confirmation from the ground.

The Anakena mosaic will be used to study the state of the archaeology at Anakena beach, and with repeated photography in future seasons, changing geological conditions and the state of archaeological structures can be studied. Similar information will also be gathered from the other ahu photographed. The ahu on the northwest coast was previously unmapped because of its large size and poor condition, but kite photography has allowed for a detailed record to be made. Finally, the photographs and mosaics will be sent back to the Museo Antropológico P. Sebastián Englert in Rapa Nui to be made available to the Rapanui people for education and preservation.

FIGURE 2 - Ahu Locations

Distribution of identified ahu locations across the island.

11

CONCLUSION

Satellite image survey and kite aerial photography have proven to be extremely low impact and labor efficient ways to supplement traditional archaeological survey. Satellite imagery allows for quick work over a large area and kite aerial photography is a highly portable and robust means with which to collect more detailed data, especially for fragile structures in remote areas. A visual survey of satellite images of the island will provide the basis for a comprehensive database following ground-truthing in upcoming field seasons. Kite photography has provided detailed images of structures for further study, education and preservation.

Future work on similar projects would do well to expand upon the scope of the project considerably. As mentioned earlier, the creation of orthophotographs would increase the usefulness of the images for making measurements. Experimentation with other methods of imaging would also be advisable. Given the high sensitivity of digital sensors to near infrared (NIR), digital photographs taken with NIR filters may reveal details that would otherwise be hidden. For even tighter ground control work, a total station could be used to measure ground

FIGURE 3 – Kite Photograph Mosaic

A section of the mosaic of Anakena beach. The rectangular structure adjoining a thin circular wall is an ahu, as is the smaller structure in the upper right corner. The black rectangles on the sand are fences protecting field school excavation units during the summer of 2005.

12

target locations, instead of a GPS unit. Continuous revisions to the rig design itself are also advisable as lighter or stronger materials become more readily available or as field experience suggests revisions to the design. Given a larger budget, it would also be helpful to use a higher resolution camera, preferably with a larger, higher quality lens. All of these improvements will take considerable experimentation to implement, but it is the author’s hope that with continued work with kite photography it will grow in its usefulness as a counterpart to satellite images in low impact survey work.

ACKNOWLEDGMENTS

The author would like to especially thank the Hawai’i Space Grant Consortium for an

opportunity to conduct research as an undergraduate and for their continuous moral and financial support, as well as NASA for it’s continuing investment in education that has been the root of this project. Mentor Dr. Terry Hunt has been very supportive and has provided the author with innumerable opportunities over the last few years. California State Long Beach’s Dr. Carl Lipo has made essential contributions to this project including satellite images, survey GPS units and considerable technical input. The financial support of the Manoa Arts and Sciences Advisory Council and the Department of Anthropology Carol Eastman Fund were instrumental in allowing photography of the ahu during the summer of 2005. A special thanks goes out to Alex Morrison, who has been an invaluable resource for ArcGIS and Dr. Everett Wingert whose expertise on aerial photography has been most helpful. Kelley Esh has been great moral support and her interest in this project has been a consistent motivator along the way.

REFERENCES

Bradford, I., Lipo, C.P. and Hunt, T.L. (2005) The Use of Multispectral Satellite Imagery to

Map Prehistoric Agricultural Features on Rapa Nui. Proceedings of the VI International Conference on Easter Island and the Pacific. Viña del Mar: Easter Island Foundation.

Cristino, C., Vargas P. and Izaurieta R. (1981). Atlas Arqueológico de Isla de Pascua. Santiago:

Corporación Toesca. Dersch, H. (2005). Panorama Tools. Retrieved January 2005 from http://webuser.fh-

furtwangen.de/~dersch/. Haefner, S. (2005). Kite Aerial Photography: Picavet Suspension. Retrieved February 2005

from http://scotthaefner.com/kap/equipment/?page=picavet. Lipo, C. P. and Hunt, T. L. (2005). Mapping prehistoric statue roads on Easter Island. Antiquity

79, 158-168. Martinsson-Wallin, H. (1994). Ahu – The Ceremonial Stone Structures of Easter Island.

Uppsala: Societas Archaeologica Upsaliensis. Wingert, E. (2005). GEOG 370 Map and Aerial Photo Interpretation. Course at the University of

Hawai'i at Manoa, Department of Geography, Spring 2005.

13

PLANNING OF KEPLERIAN ORBITS: APPLICATION TO PERIODIC TRAJECTORIES

Vann Michael Bennett

Department of Mathematics University of Hawai’i at Manoa

Honolulu, HI 96822

ABSTRACT Satellites have numerous uses ranging from the transmission of communications to the

observation of the earth’s weather. Depending on the function of a given satellite, the satellite will have a particular orbit that suits it best. The particular function of a satellite may sometimes change, thereby requiring a change in its orbit. Careful planning is a necessity in the alteration of any satellite’s orbit. The complex nature of this planning requires that it be done through both purely analytical techniques and through approximating numerical techniques before the transfer of the orbits. In this project, implications are found on certain orbital parameters of a periodic trajectory after making reasonable assumptions on some of the other orbital parameters; thereby constructing a periodic trajectory. This newly constructed periodic trajectory is then analyzed using the theory of time-reversal symmetry in dynamical systems. The goal of this analysis is to find a more general method of constructing periodic trajectories and to hopefully make new observations on the interconnectedness of the various orbital parameters. Visual aides, in the form of plots of satellite position versus time, provide an invaluable picture of exactly what the focus of this project is. These plots are obtained using the numerical computing software of MATLAB along with the pertinent Keplerian equations of motion and a few arbitrary initial conditions.

INTRODUCTION

On October 4, 1957, the U.S.S.R. launched Sputnik 1, and became the first nation to put

an artificial satellite into orbit around the earth. This is considered by many to be the height of the Cold War and the gun at the starting line of the Space Race. The U.S.A. answered shortly thereafter with Explorer 1, and for the next few years the two superpowers stayed neck and neck, with the U.S.S.R. a hair’s width in the lead. The American Alan Shepard was in fact the second human to enter space by a mere 23 days, having been beaten by the cosmonaut Yuri Gagarin, who crossed the threshold on April 12, 1961. However, the U.S.S.R. was overtaken on July 21, 1969 when Neil Armstrong stepped from Apollo 11 onto the surface of the moon, marking a defining moment of the 20th Century as well as the culmination of the Space Race.

Today, satellites are nothing more than a mundane reality of technologically advanced nations. Their functions include communications, earth observation, weather observation, navigation, and reconnaissance. This spectrum of functions requires a spectrum of orbits, or paths that satellites follow. Communications satellites must generally maintain a fixed position relative to the earth in order to effectively relay signals, this is achieved through following what is known as a geosynchronous orbit. Many weather observation satellites collect information concerning the entire planet twice each day by following sun synchronous polar orbits. Satellites are launched with powerful rockets into space, but it is difficult to launch them precisely into

14

their desired orbits. They must therefore have means of propulsion to obtain their correct paths. Omnipresent perturbations also make means of propulsion necessary for proper satellite function. The propulsion mechanism must obviously be taken into account in the planning of an orbit. This is done with the sometimes unwieldy 3D controlled Kepler equations of motion. Approximate numerical solutions to this equation obtained using MATLAB greatly reduce the time needed for planning, and hence simplify the entire process.

The goal of this project is to attack the problem of planning Keplerian orbits from just a single front. The front considered here is that of taking an initial trajectory on a given time interval and creating a periodic orbit that follows this initial trajectory for exactly half of the orbit. To simplify the problem without losing significant generality, motion is considered in only one plane. Concepts from the theory of time-reversal symmetry are then closely examined in order to determine their connectedness to this project.

3D CONTROLLED KEPLER EQUATIONS OF MOTION

The following system of differential equations together make up what are known as the

3D controlled Kepler equations of motion:

)()(3

10 xfxf

dtxdx i

ii

∑

=

+== γ ; where

=

LhheeP

x

y

x

y

x

;

=

PW

Pf

2

00

00000

µ;

−=

000

cossin0

01

LL

Pfµ

;

++++

=

000

)sin(sin)cos(cos

2

02

WLeLWLeL

WP

Pf y

x

µ

;

−

=

Z

LC

LCZe

Ze

PW

fx

y

sin2

cos2

0

1

03 µ

; and

++=

−=

++=

221

cossin

sincos1

yx

yx

yx

hhC

LhLhZLeLeW

.

15

In this equation x is a function of time that represents the motion of a given satellite in terms of the Gaussian coordinates P , xe , ye , xh , yh , and L . A Keplerian orbit is simply an orbit that obeys these equations. Gaussian coordinates are a convenient route for describing ellipses in three dimensions. P (t) is the length of the semi-latus rectum and hence the size of the ellipse at time t, xe (t) and ye (t) together represent the eccentricity and hence the shape of the ellipse at time t, xh (t) and yh (t) represent the inclination of the ellipse with respect to three mutually orthogonal constant coordinate axes, and L (t) is the angular position of the satellite with respect to the aforementioned axes at time t. 1γ (t), 2γ (t), and 3γ (t) represent the propulsion mechanism of the satellite with each function being the thrust of the satellite in one of three mutually orthogonal directions at time t; one direction is tangent to the satellite’s path, one is perpendicular to the path and in the orbital plane, and the third is perpendicular to each of the others. Since motion in only one plane is considered, it is taken that 3γ (t) ≡0. Making the previous assumption and setting equal corresponding components of the vectors involved in the 3D controlled Kepler equations of motion gives the following somewhat simpler system:

2

2/3

0

2 γµ W

PP =;

)]cos

(cossin[ 210 W

LeLLPe x

x+

++= γγµ

;

)]sin

(sincos[ 210 W

LeLLPe y

y

+++−= γγ

µ ;

2/3

2

0 PWL µ= .

Solutions of this system of nonlinear differential equations describe the planar orbit

followed by a satellite with a thrust of 1γ (t) and 2γ (t). Note that this simplified planar case of the 3D controlled Kepler equations does not include xh and yh anywhere.

TIME-REVERSAL SYMMETRY IN DYNAMICAL SYSTEMS

In the theory of time-reversal symmetry in dynamical systems, a function R is said to be a

reversing symmetry if ))(()( xRFdt

xRd

−= , where x represents a trajectory, R represents a

transformation on the trajectory, and F

represents the applicable forces on the trajectory. It is important to note that nowhere in this equation is a control function, which this problem requires. The following function R, which acts on x as well as the control functions, meets the above criteria for being a reversing symmetry:

16

T

yxyx tttLtetetPtttLtetetPR ))(),(),(),(),(),(())(),(),(),(),(),(( 2121 −−−−−−−−−−= γγγγ

. Observe that

Tyx

yx tttLtetetPdt

tttLtetetPRd))(),(),(),(),(),((

))(),(),(),(),(),((21

21 −−−−−−−−−= γγγγ

)),(),(),(),(),(),((())(),(),(),(),(),(( 2121 tttLtetetPRFtttLtetetP yx

Tyx γγγγ

−=−−−−=

assuming that the following conditions hold: P (t) = P (-t), xe (t) = xe (-t), ye (t) = - ye (-t),

L (t) = - L (-t), 1γ (t) = 1γ (-t), and 2γ (t) = - 2γ (-t). Thus R would be a reversing symmetry if it did not act on the control functions, for this reason it will be called a time-reversal symmetry for a control system.

SYMMETRY AND PERIODICITY

Given an arbitrary planar trajectory denoted by the Gaussian parameters P , xe , ye , and

L with control functions 1γ and 2γ on the time interval from t = -T to t = 0, a method of finding a periodic orbit that follows this trajectory for half of the orbit is desired. The problem is to find functions P , xe , ye , L , 1γ , and 2γ on the time interval from t = 0 to t = T such that each of the pertinent Gaussian parameter functions is continuous at all points in the interval from t = -T to t = T and directly from t = T to t = -T, and that the simplified planar case of the 3D controlled Kepler equations is satisfied. Observe that these condition are met by applying the function R to the initial trajectory. The requirement that each of the Gaussian parameter functions have the same value at t = -T and t = T maintains the continuity of the satellite motion. This continuity condition requires that ye (T) = - ye (-T) = 0 and L (T) = L (-T) = 0. Furthermore, it is taken that

0≡ye , for ease of handling the equations. Satellites cannot jump through space, and this insures that any solutions that do jump through space are eliminated. Any function f such that f(x) = f(-x) is said to be even; the graphs of even functions are symmetric about the axis of the dependent variable, in this case the f axis. Assuming that 0≡ye allows for the computation of

1γ in terms of 2γ , by the third equation derived from the simplified planar case of the 3D controlled Kepler equations. It can be shown in a straightforward manner that

)11(tan21 WL += γγ . This greatly simplifies the numerical integration of the system, which is

performed in the following section.

APPLICATION AND DEMONSTRATION

Given an arbitrary “half” orbit on an interval [-T,0], the methods described above describe how to return to the initial configuration on the time interval [0,T], and to repeat the

17

cycle indefinitely. An arbitrary set of data that could apply to many modern day satellites was chosen to demonstrate visually the implications of the function R above. The 3D controlled Kepler equations were integrated using MATLAB and the arbitrary set of data was used for the initial conditions of the various differential equations. Figure 1 is the initial “half” orbit starting at time t = -T and ending at time t = 0. The object located inside of the trajectories represents the earth. One unit on each figure represents 10 megameters, or 10^7 meters. The function R was then applied to this “half” orbit in order to obtain Figure 2, starting at time t = 0 and ending at time t = T. Note that the end of the path in Figure 1 is continuous with the beginning of the path in Figure 2, and that indeed the orbit is periodic.

Figure 1: the initial “half” path. Figure 2: the calculated return trajectory.

CONCLUSION

Perhaps the most important part of this project is not the discovery of the time-reversal symmetry R for this control system, but the interpretation of R. Figure 1 above is a concrete example of an arbitrary “half” path. However it could be any continuous path and the application of R to the Gaussian parameter functions would yield the symmetric path that completed the orbit into a periodic one.

This technique could prove to be quite useful in the planning of Keplerian orbits. The fact that nearly any path could be followed and subsequently “completed” into a periodic orbit has many implications. It is impossible to imagine the number of reasons why a satellite might be required to follow a particular path. But with this reversal-symmetry R for this control system, a methodical way of deciding what to do after the path is followed has been determined.

The implications of the time-reversal symmetry R for this control system however are quite limited. The simplification of the three-dimensional problem into a two-dimensional problem is significant and should not be overlooked. Very rarely would a satellite require planning in a single plane. The elimination of the third control function 3γ greatly simplified the system of nonlinear differential equations, particularly by eliminating xh and yh from the pertinent system. If 3γ were again considered, the difficulty in solving the 3D controlled Kepler equations would grow exponentially. Not only would each of the differential equations present

18

now be complicated with the addition of xh and yh , but two entirely new differential equations would appear, those being the two associated with the derivatives of xh and yh .

ACKNOWLEDGEMENTS

The author would like to thank his mentor Dr. Monique Chyba along with Dr. Thomas

Haberkorn, who was vital for the numerical integration with MATLAB. The author would also like to thank the Hawai’i Space Grant Consortium for giving him this opportunity to experience the research process.

REFERENCES

Hairer, E. (2004) Solving Ordinary Differential Equations, Springer Verlag, Berlin.

Lamb, J., Roberts, J. (1997) Time-reversal symmetry in dynamical systems: a survey.

Physica D.

Pollard, H. (1966) Mathematical Introduction to Celestial Mechanics, Prentice-Hall, Inc., New Jersey.

Sontag, E. (1998) Mathematical Control Theory: Deterministic Finite-Dimensional Systems, Springer Verlag, Sydney.

19

RECONSTRUCTION OF THE EVOLUTION OF GALAXIES USING THE EXTENDED-PRESS-SCHECHTER MODEL WITH

GLOBULAR CLUSTERS

Eli Bressert Department of Physics and Astronomy

University of Hawaii at Hilo Hilo, HI 96720

ABSTRACT

Galaxies are large gravitationally bound systems that consist mainly of stars, interstellar

gas and dust, and dark matter. The Milky Way, our own galaxy and home to Earth, has evolved from its infantile stage billions of years ago to what we see today. A question one may pose is how did the Milky Way form and is its evolution different to those of its neighboring galaxies. In the general case galaxy evolution can be depicted through collision and cannibalization, but to observe a galaxy and know its history has proven to be a difficult task. In this research a solution to the withstanding problem via a new method will be discussed. By using an advanced semi-analytical model known as the extended-Press-Schechter method and globular clusters we have obtained Monte Carlo data that resembles observed galaxies, whose implications are discussed in the paper.

1. INTRODUCTION

Among the problems that still persist in astronomy, the origin and evolution of galaxies remain elusive. Since the time that Edwin Hubble discovered galaxies nearly a century ago, tracing the evolutionary path of these objects has proven to be an ambitious venture. The primary questions concerning galaxy origins are the following

"1) From a homogeneous universe how did the inhomogeneous one we see today come about? 2) How did galaxies form in the early stages of the Universe? 3) How do galaxies change over time?" 1

While all these questions have been thoroughly researched, a new method has been brought forth that applies particularly to the third question by using properties of globular clusters to determine galaxy histories. Globular clusters are dense gravitationally bound objects that contain anywhere between 104 to 106 stars which orbit around galaxies. An example of a globular cluster is shown in Figure 1. An interesting and potentially useful property of globular clusters is the bimodality of its metallicity distribution.

1 http://en.wikipedia.org/wiki/Galaxy_formation_and_evolution

Fig. 1. Globular cluster M80 observed on the Hubble Telescope. Image Credit: The The Hubble Heritage Team (AURA/STScI/NASA)

20

Why would one use the bimodal property of globular clusters to describe galaxy histories? Let us note some well-known facts of globular clusters to answer the question. First, globular clusters are of ancient origin dating back to 10 to 15 billion years ago (West et al. 2004). This means that the globular clusters were companions of the birthing galaxies and have remained unscathed through galaxy evolution because of the relatively large gravitational attraction between the stars. Second, the number of globular clusters per galaxy correlates with the luminosity of their parenting galaxies (West et al. 2004). As galaxies undergo evolution we have found compelling evidence of numerous events of galaxy mergers or larger galaxies "cannibalizing" smaller galaxies. With the large perturbations that are involved in galaxy evolution the initial galactic structures will have undergone change. What may been initially a spiral galaxy could turn into an elliptical galaxy after merging with its neighboring galaxies. The globular clusters of the galaxies during these merging and cannibalizing events remain unperturbed. Hence, it may be possible to use these globular clusters as 'fossils' (West et al. 2004) to determine the unique histories of galaxies. In this paper we will study only the history of elliptical galaxies. To test the idea that has been put forth by Cote, West, and Marzke (1993) Monte Carlo simulations are used. There are two ways to undertake this problem: (1) N-body model or (2) the Press-Schechter model. Here we focus on the Press-Schechter formalism because of its advantages that are described in the next section. An astronomer from the Astronomical Observatory of Rome, Dr. Nicola Menci, has collaborated and provided a program2 utilizing the extended-Press-Schechter method (EPS), which produce galaxy histories. To test the EPS program this thesis extends Menci's simulations by adding the globular clusters, which allows us to study detailed components of galaxy evolution. With direction provided by Dr. West and Dr. Marianne Takamiya I have written a program that allows us to study globular cluster population in the galaxies simulated by Dr. Menci's EPS program.

2. MONTE CARLO MODELS

Recently there have been large advancements in simulations of galaxy evolution. There are two primary tracks that one can embark to create Monte Carlo simulations. The first, N-body model, is the most widely-used and popular method where sets of differential equations are solved to track motions of bodies that interact with one another over a set period of time. The other method known as the Press-Schechter model was created by two Caltech graduate students in 1974 to bypass the computational difficulties that come along with the N-body model (Press et al.). Brief descriptions of the two types of simulations are presented next, to understand the reason why the Press-Schechter model is implemented in this research.

2.1 N-BODY METHOD The N-body method is a flexible and modifiable system. Ranging from particle physicists

to astronomers, they use the model to solve a diverse number of problems encompassing the atomic to the cosmological scale. In each of these domains different assumptions have to be made to allow the N-body simulations to work properly. For example an astronomer studying stellar interactions will devote their research to the interactions between the particles in terms of collisions. On the other hand, an astronomer working on galaxy evolution prefers collision-less interactions because collisions between stars are negligible (Binney, 1987). Due to the intensive computations of N-body simulations at the galactic scale, the stars are clustered into 2 http://lbc.mporzio.astro.it/menci/

21

superparticles (Bertin, 2000). In one case of galaxy evolution simulations, 100 million particles are used to represent the Milky Way. Using the N-body model allows the user to obtain comprehensive and accurate data for galaxy evolution simulations, but at the cost of intensive supercomputer usage with extended periods of time. For example, a simulation representing the Milky Way and Andromeda galaxies colliding took a period of around four3 days to complete.

Another problem that N-body simulations pose is the particle resolution. At galactic scales in these simulations the superparticles represent a large collection of stars that contain 100 times more particles than globular clusters. This means that in a typical N-body model at the galactic scale one would not be able to track the dynamics of the globular clusters. The study of globular cluster systems as proxies for galaxy evolution is cumbersome using N-body simulations because higher mass resolution demands unrealistically large supercomputer power and time.

2.2 EXTENDED-PRESS-SCHECHTER METHOD

The original Press-Schechter model has been used since its inception in 1974 in numerous studies. Over the span of the last 30 years the model has been modified to provide increasingly accurate solutions. Two of the currently modified forms are the extended-Press-Schechter (Lacey et al. 1993) and modified-Press-Schechter (Salvador-Sole et al. 1998) models. The milestone of the Press-Schechter model in general is its ability to bypass the non-linearity of the N-body model by making statistical assumptions. The upshot of employing the EPS model is its ability to be executed on a personal computer and requiring relatively minimal computer time. Having the ability to run galaxy simulations many times on a regular computer allows us to re-parameterize the system for optimal performance. EPS supersedes high-resolution N-body simulation models. Therefore, the EPS simulation method is efficiently the best choice in terms of time, computing power, and flexibility for this research.

3. MENCI’S PROGRAM

The EPS program from Menci's research provides the merging trees of several galaxies. We studied the merging histories of 1100 galaxies. Menci’s program (hereafter MP) output consists of redshift, total dark matter mass, total stellar mass, and final absolute visual magnitude of the galaxies involved. The typical format of the output is an X by 5 matrix where X is determined by the number of galaxies and interactions that are present in the output file, which we can refer to as the merger tree. In the research the current output contains roughly 1,100 unique merger trees for final galaxies.

4. GLOBULAR CLUSTER PROGRAM

With MP's output being a record of multiple galaxy merger trees we require a program that assigns globular clusters to galaxies that appears for the first time in the merger trees. Since there was no program of this sort readily available in the public domain I wrote a program in IDL4 to do the required task. We can refer to the globular cluster program as GCP. The GCP algorithm is as follows: 3 http://www.newsandevents.utoronto.ca/bin/000414b.asp 4 Interactive Data Language - http://rsinc.com/idl/

22

1. Search for young galaxies and assign globular clusters accordingly. The globular clusters' mean metallicity values are determined by

Fe /H[ ]= a0 + a1MVi + a2MV

i2

where

a0,

a1, and

a2 are constants and

MVi is the absolute visual magnitude of the

corresponding galaxy (Cote et al. 1993). 2. Determine the number of globular clusters per galaxy from the specific frequency. The

specific frequency (

SN ) is defined as the number of globular clusters per galaxy luminosity. Based on McLaughlin's equations (McLaughlin, 1999) we ascertain the specific frequency, which is dependent on the galaxies' absolute visual luminosity, denoted as

LV . For galaxies with

LV < 3.5 ×109L solar, where

L solar is the total luminosity unit of the Sun, we have

SN = 5 LV

2 ×108L solar

−0.4

For galaxies with

LV > 3.5 ×109L solar we use

SN = 5 LV

3 ×1010L solar

0.75

With the specific frequencies defined for the galaxies we use the following equation and solve for

NT :

SN = NT ×100.4 M V +15( )

By solving for

NT we obtain the number of globular clusters per galaxy as a function of their galaxies’ luminosities.

3. Sum up the number of globular clusters in each final galaxy. To test the validity of the results of the simulation we compare metallicity distribution of the galaxies for simulations vs. observations. A sample of simulation data vs. observation is provided in a histogram format, which can be seen in Figure 2.

5. KMM TESTING

The data from the globular cluster program is best analyzed in histogram format. In these histograms we look for bimodality. In some cases the mean values of the curves are indeterminate from one another visually. Hence, using a statistical algorithm software known KMM5 (Ashman, Bird & Zepf 1994) allows us to test the difficult data sets for bimodality.

5 http://cas.umkc.edu/physics/ashman/blake.html

23

6. RESULTS

GCP has various options for computing the luminosity of the galaxies based on the stellar matter mass provided in MP through the mass-luminosity ratio. Since we are dealing with multiple different galaxies in sequential time the mass-luminosity ratio contains a large range of values. Hence, we use mass-luminosity ratios that best fit the galaxies in the merger trees. The results from the simulations provide globular cluster populations of similar characteristics to observed data. In addition, we apply the KMM test to the simulated results to obtain verification of bimodality and the mean values of the peaks. An example of this can be seen in Figure 2.

7. FUTURE

As the Monte Carlo simulations are tested and more results are produced we will use one more statistical method to test for the validity of the data; the Kolmogorow-Smirnov method. This method is used to test goodness-of-fit between the simulated and observed data of galaxies

Fig. 2. The simulated galaxy, on top, which has MV = −22.25 is a plot that represents a simulated Monte Carlo globular cluster set that is comparable to M87’s globular cluster set. The number of globular clusters for the simulated galaxy and M87 are approximately the same at ~1500. KMM testing of the simulation data gives strong indication that it is bimodal with its mean peak values at -1.52 and -0.80.

24

and their respective globular cluster population characteristics. In the GCP there are placeholders for more parameters to be added for more complex assumptions concerning globular cluster formation. These placeholders will most likely be used for events such as galaxy mergers where stellar gases are exposed to density perturbations from the collisions and new globular clusters are formed. Another possible assumption that will be appended is the globular clusters’ mean metallicity as a function of redshift. By working from the simplest assumptions we can append the more complex parameters for better precision in the simulations. Last but not least, MP is being configured to compute precise values of absolute visual magnitudes of galaxies in the merger tree outputs. This will provide GCP the ability to produce more accurate Monte Carlo results of globular clusters populations.

AKNOWLEDGEMENTS

I would like to thank Dr. Michael West for allowing me to help him on this research. The experienced gained in the numerical and theoretical aspects is important and has re-affirmed my chosen direction in the astronomical field. His time and efforts has helped this fellowship move forward tremendously. Also, many thanks to Dr. Marianne Takamiya who has helped me in the final stages of this research. Her critical and sharp questions and steadfast support has made the research’s dark corners lighten up. The NASA Space Grant Consortium who has funded the fellowship, always provided positive feedback on the research. Last but not least, I would like to thank my wife Judith van Raalten who has provided thorough questions and excellent editing skills to the research and papers.

25

REFERENCES

Ashman, K.M., Bird, K.A., Zepf, S.E. (1994) Detecting bimodality in astronomical datasets. AJ 108, 2348.

Bertin, G. (2000) Dynamics of Galaxies, Cambridge University Press, Cambridge, UK. Binney, J., Tremaine, S. (1987) Galactic Dynamics, Princeton University Press Princeton.

Cambridge, UK. Cote, P., Marzke, R.O. & West, M.J. (1998) The formation of giant elliptical galaxies and their

globular cluster systems. AJ 501, 554-570. Harris, W. E., & van den Bergh, S. (1981) Globular clusters in galaxies beyond the local group.

I - New cluster systems in selected northern ellipticals. AJ 86, 1627. Lacey, C., Cole, S. (1993) Merger rates in hierarchal models of galaxy formation.

MNRAS 262, 627. McLaughlin D. (1999) The efficiency of globular cluster formation. AJ 117, 2398. Press, W. H., & Schechter, P. (1974) Formation of galaxies and clusters of galaxies by

selfsimilar gravitational condensation. ApJ 187, 425 Salvador-Sole E., Solanes J. M., Manrique A. (1998) Merger vs. accretion and the structure of

dark matter halos. ApJ 499, 542. West, M.J., Cote, P., Marske, R.O., & Jordan, A. (2004) Reconstructing galaxy histories from

globular clusters. Nature 427, 31.

26

FIRST INFRARED SPECTROSCOPIC CHARACTERIZATION OF THE Ge2H(3,5) MOLECULES AND Ge2D(3,5) RADICALS IN LOW TEMPERATURE

GERMANE MATRICES

William Carrier Department of Chemistry

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

The digermyl, Ge2H5(X2A’) its D5-isotopomer, Ge2D5(X2A’) radicals and the digermenyl radical, H2GeGeH(X2A”), its D3-isotopomer D2GeGeD(X2A”) were observed for the first time in low temperature germane and D4-germane matrices at 12 K via infrared spectroscopy upon irradiation of the low temperature ices with energetic electrons. The most intense digermyl absorption the ν6 fundamental was observed at 765 cm-1and 561 cm-1 for Ge2H5 and Ge2D5, respectively. The digermenyl ν3 fundamentals were observed at 1825 cm-1

Ge2H3(X2A”) and 1317 cm-1 Ge2D3(X2A”), respectively. Our investigations suggest that this radical is formed via radiolysis of germylgermylene, H3GeGeH(X1A'), and digermene, H2GeGeH2(X1Ag). Experimental results for the ν5 fundamental verified also the first observation of the H3GeGeH(X1A') isomer at 780 cm-1 and of 558 cm-1 for its deuterated counterpart. We also assigned the ν11 fundamental for H2GeGeH2(X1Ag) as observed at 845cm-

1 and the ν5 mode of D2GeGeD2 at 1475cm-1. The infrared absorptions of the hitherto elusive Ge2H5(X2A’) and H2GeGeH(X2A”) radical may aid in monitoring chemical vapor deposition processes via time resolved infrared spectroscopy of germane and can provide vital guidance to search for this hitherto undetected germanium-bearing molecule in the atmospheres of Jupiter and Saturn.

INTRODUCTION A detailed understanding of the structures, spectroscopic properties, and energetics of simple hydrogenated germanium clusters of the generic formula Ge2Hx (x=1-6) are of growing importance to technological applications such as germane chemical vapor deposition (CVD) and semiconductor processing.1,2 Plasma etching processes, reactive plasma, and chemical vapor deposition techniques are of wide technological interest to produce germanium-bearing nano particles and amorphous germanium films (a-Ge:H) via microelectronic engineering.3 Germanium films also have applications in material sciences, particularly in the development of solar cells, electro-photographic drums, and arrays for liquid crystal displays. In chemical vapor deposition processes, germane (GeH4) or digermane (Ge2H6) precursor molecules are decomposed by hot filaments or electrons generated within plasmas.4 During this process, germanium bearing species such as GeHx (x=1-3) and dinuclear clusters like Ge2Hx (x = 1 - 5) have been identified in the gas phase as major growth species to produce amorphous, often porous germanium films. The germane molecule – a potential precursor to complex organo germanium molecules - has been identified with relative abundance of 7x10-10 and 4x10-10 relative to molecular hydrogen in Jupiter and Saturn, respectively.1 A photolysis of germane could form highly reactive GeH3

27

and GeH2 radicals, which could react subsequently to form hitherto unobserved Ge2Hx species. An identification of these species will provide a better understanding to the origin and evolution of the atmospheres of the giant gas planets. Germane has also been observed in the dying carbon star IRC+10216. Note that germane CVD processes modeled via complex reaction networks also require an in depth knowledge of the underlying chemical mechanisms. Currently, in situ characterization of gaseous molecules in CVD processes is predominantly carried out via mass spectrometry,5 but this technique can hardly discriminate between structural isomers; furthermore, lacking the fragmentation pattern of these germanium-bearing molecules makes it difficult to observe the corresponding radicals. To provide input to CVD reaction networks, a systematic study of the thermochemical properties and adiabatic ionization potentials of the germanium hydrides was performed very recently.6 However, despite the importance of the Ge2Hx (x=1-6) species in chemical vapor deposition processes of germane, no time resolved spectroscopic probes such as infrared spectroscopy have been established – predominantly because only limited information on the infrared absorptions of these species are available. Here, a few vibrational spectra of Ge2Hx (x = 2,4,6) have been reported;7,8,9,10 the most intense infrared absorptions of digermane, Ge2H6(X1A1g) [ν6 and ν8; 752 cm-1 and 881 cm-1], digermene, Ge2H4(X1Ag) [ν11; 789 cm-1], and the di-bridged Ge2H2(X1A1) [ν6; 972 cm-1] have been experimentally observed in low temperature neon matrices.(15) Although the Ge2H4 absorption was suggested by the authors to be assigned to ν11 of the Ge2H4(X1Ag) isomer, our calculations indicate that this is more likely to identify this as belonging to ν5 of the Ge2H4(X1A’) structure. Surprisingly, data on the corresponding radical species (x = 1,3,5) have not previously been recorded.

EXPERIMENTAL

The experiments were performed in a contamination-free ultrahigh vacuum (UHV) machine.11 Its main chamber can be evacuated down to 5 × 10−11 torr by a magnetically suspended turbo pump backed by an oil-free scroll pump. A two stage closed cycle helium refrigerator; connected to a differentially pumped rotary feed through, is attached to the main chamber and holds a polished silver mirror. The silver mirror can be cooled to 10 K and serves as a substrate for the ice condensate. The gas samples can be brought into the chamber through a precision leak valve, which is connected to a gas reservoir and supported by a linear transfer mechanism. The deposition system can be moved 5 mm in front of the silver mirror previous to gas condensation. This setup guarantees a reproducible thickness of the ice samples. The germane ices were prepa-red at 12 K by depositing germane (99.99%) and D4-germane (99.99%) at pressures of about 7 × 10−8 torr for 30 min onto the cooled silver mirror. These ices were irradiated for 60 min by scanning the ice over an area of 3.0 ± 0.4 cm2. The irradiation was conducted at 12K with 5 keV electrons generated via an electron gun at beam currents of 10 nA, 100 nA and 1000 nA. Background analysis was performed by collecting data with no germane in the UHV chamber.To guarantee an identification of the reaction products in the solid state, a Fourier transform infrared spectrometer was used. The Nicolet 510 DX FTIR unit (5000-500cm-1) operated in an absorption-reflection-absorption mode (reflection angle α = 75º) at a resolution of 2 cm-1. The infrared beam was coupled via a mirror flipper outside the spectrometer, passed through a differentially pumped potassium bromide (KBr) window, was attenuated in the ice sample prior and after reflection at a polished silver waver, and exited the main chamber through a second

28

differentially pumped KBr window before being monitored via a liquid nitrogen cooled detector. The gas phase is monitored via the Balzer QMG 420 (1-200 amu mass range) operating with electron impact ionization of the neutral molecules in the residual gas analyzer mode at electron energies of 100 eV and the photomultiplier running at 2000 V. The dwell time of the masses was chosen to be 0.5 ms. Comparing our data with previous literature suggests a germane phase III in our experiment. Spectroscopic observation of solid germane in low temperatures crystals reveals α, β, and γ lattice modes.12 It is fairly easy to discriminate phases I – IV of germane via infrared spectroscopy of the ν1 - ν3 and ν2 - ν4 regions.13 To determine the thickness of the sample we integrated the infrared absorption features at 2111 and 821 cm-1. The ice thickness was then calculated using the Lambert-Beer relationship.14 The integrated absorption coefficients of these fundamentals, 5.5 × 10-17 and 4.7 × 10-17 cm mol-1 respectively and the density of the germane ice 1.751 g cm-3 15 determined an optical thickness of 54 ± 20 nm.

4500 4000 3500 3000 2500 2000 1500 1000 5000.0

0.1

0.2

0.3

0.4

0.5

0.6

abso

rption

wavenumber, cm-1

Figure 1: Infrared spectrum of germane frost at 11K

Table 1: Infrared absorptions of the germane (left column) and d4-germane (center column) frosts (sh: shoulder); α, β, and γ denote lattice modes of the germane sample.

frequency, cm-1 frequency, cm-1 assignment

4190 3002 2 ν3

4120 2889 ν1 + ν3

3025 2172 ν2 + ν3 + α

3000 2156 ν2 + ν3

29

frequency, cm-1 frequency, cm-1 assignment

2200 1597 ν3 + γ

2140 1543 ν3 + β

2114 1524 ν3 + α

1737 1247 ν2 + ν4 + α

1715 1231 ν2 + ν4

958 681 ν4 + γ

914 652 ν2

846 616 ν4 + β

826 598 ν4 + α

790 574 ν4

THEORHETICAL APPROACH The molecular structures and vibrational frequencies of the GeHx species (x=1-4) as well as digermane (Ge2H6), digermyl (Ge2H5), germylgermylene (H3GeGeH), digermene H2GeGeH2(X1Ag) and digermenyl, H2GeGeH(X2A”) radicals together with their per-deuterated were scrutinized in terms of ab initio molecular orbital methods. The geometries were optimized with the hybrid density functional B3LYP method, i.e. Becke’s three-parameter non-local exchange functional16 with the non-local correlation functional of Lee, Yang, and Parr 17 and the 6-311G(d,p) basis set.18 Since the energy of the quartet electronic state of the Ge2H5 radicals are extremely high and dissociative, we have only studied the doublet electronic state. We have examined various isomeric structures of Ge2H5 species for mono-bridged, di-bridged, and tri-bridged geometries in addition to the H3GeGeH2 structure in analogy to the Si2H5 system.19 All optimized geometries with di-bridged and tri-bridged structures are shown to have more than two imaginary frequencies and they turned into either two structures when we optimized further along each imaginary mode. The vibrational frequencies and infrared intensities were obtained for optimized GeHx (x = 1 - 4) and Ge2Hx (x = 3,4,5, 6) species and their deuterated isotopomers. The coupled cluster CCSD(T) calculations20,21 with the aug-cc-pVTZ basis set22 were also conducted at the optimized structures obtained with the B3LYP method in order to compare the relative energies for the systems. All computations were carried out using the GAUSSIAN 98 program package.23 The relative energies stated in the text are the values obtained with the CCSD(T) method corrected with the zero-point vibrational energies obtained with the B3LYP method.

EXPERIMENTAL RESULTS Infrared absorptions of the germyl radical, GeH3(X2A1), appeared instantaneously with

the start of the irradiation of the germane sample with an electron current of 10 nA at 12K at 665 cm-1. Note that the absorptions of GeH(X2Π) and GeH2(X1A1) species were not detected in our

30

investigation. Features of the digermane molecule, Ge2H6(X1A1g), also appeared immediately after the initiation of the irradiation (Figure 6). Here, we were able to observe an absorption at 752 cm-1; this can be assigned to the ν6 umbrella mode. Having identified the germyl radical and the digermane molecule, we exposed the germane matrix to an enhanced irradiation time. The primary objective of this approach is to examine if the digermane molecule can also undergo a germanium-hydrogen bond rupture yielding the digermyl radical, Ge2H5(X2A’). As a matter of fact, we were able to detect an additional absorption at 765 cm-1.

Table 2. Newly observed species and their absorptions in low temperature germane matrices.

species frequency, cm-1

fundamental species frequency, cm-1

fundamental

GeH3 665 ν2 GeD3 608 ν4

Ge2H6 870 ν11 Ge2D6 626 ν11 752 ν6 529 ν6

Ge2H5 765 ν6 Ge2D5 616 ν4/ ν12 561 ν6

Further investigation of spectroscopic data identified both germylgermylene (H3GeGeH) and digermene H2GeGeH2(X1Ag) absorptions at 785 and 845 cm-1, respectively. Further hydrogen bond rupture was determined with the digermenyl ν3 fundamental observed at 1825 cm-1

Ge2H3(X2A”). The end of the suggested formation route was concluded with an absorption at 792 cm-1 representing the Ge2H2(X1A) radical.

Table 3. Newly observed species and their absorptions in low temperature germane matrices.

species frequency, cm-1

fundamental species frequency, cm-1

fundamental

Ge2H4 845 ν11 Ge2D4 1475 ν5

H3GeGeH

785 ν5 D3GeGeD 558 ν5

Ge2H3 1825 ν3 Ge2D3 1317 ν3

Ge2H2 792 ν2 Ge2D2 570 ν2

DISCUSSION AND SUMMARY The infrared and mass spectroscopic data imply that the response of the germane ices upon the electron irradiation is governed by an initial germane – hydrogen single bond rupture which forms atomic hydrogen plus the germyl radical, GeH3(X2A1). In order to escape the

31

[GeH3…H] matrix cage, the hydrogen atom needs an excess energy – the lattice binding energy -

of a few tens of kJmol-1 to escape.24 Otherwise, atomic hydrogen will recombine with the germyl radical to react back to a germane molecule. The infrared data suggest that the digermane molecule is formed initially within a single trajectory of the electron. Considering the unit cell of a germane crystal,21 about one germyl radical is generated on average per unit cell. Here, neigh-boring germyl radical can recombine to the digermane molecule Ge2H6(X1A1g) at 12 K if they have the correct recombination geometry. Summarized, the digermyl, Ge2H5(X2A’) and d5-digermyl, Ge2D5(X2A’), radicals were detected for the first time in germane and d4-germane matrices at 12 K via infrared spectroscopy.

Further radiolysis of the germane matrix was investigated using ab initio calculations for infrared frequencies of germylgermylene (H3GeGeH), digermene H2GeGeH2(X1Ag) and digermenyl, H2GeGeH(X2A”) radicals. We analyzed the temporal evolution for eight determined reaction formations using thirteen ordinary differential equations. The ordinary differential equations were solved using a reaction pathway program written for MATLAB. It was found from this, that the digermyl radical is indeed a decomposition product of the digermane molecule. Temporal profiles revealed that the formations of both germylgermylene (H3GeGeH) and digermene H2GeGeH2(X1Ag) isomers were formed from radiolysis of Ge2H6. Furthermore both of these isomers were found to undergo single hydrogen bond loss and form the Ge2H3 radical. That Ge2H2 was identified as the end of the formation reaction following yet another hydrogen loss from the Ge2H3 radical.

ACKNOWLEDGMENTS These experiments were supported by the NASA Space Grant administered by the University of Hawai’i at Manoa. Special thanks to Dr. Kaiser for providing mentoring throughout the entire research project. The author would also like to thank Weijun Zheng for his assistance in training on the surface scattering machine.

32

REFERENCES

1 A Ricca, CW Bauschlicher. (1999) J. Phys. Chem. A, 11121. 2 SD Chambreau, J Zhang. (2002) Chem. Phys. Lett 351, 171. 3 Special Issue ‘Surface Chemistry - Advances and Technological Impact. Chemical

Reviews 96 (1996). 4 J Perrin, M Shiratani, P Kae-Nune, H Videlot, J Jolly, J Guillon. (1998) Journal of

Vacuum Science & Technology: Vacuum, Surfaces, and Films A, 278. 5 B Ruscic, J Berkowitz. (1991) J. Chem Phys. 95, 2416. 6 Q-S Li, R-H Lu, Y Xie, HF Schaefer, III. (2002) Journal of Computational Chemistry 23,

1642-55. 7 VA Crawford, KH Rhee, MK Wilson. (1962) J. Chem. Phys. 37, 2377. 8 X Wang, L Andrews, GP Kushtu. (2002) J. Phys. Chem. A, 5809. 9 J Urban, PR Schreiner, G Vacek, PvR Schleyer, JQ Huang, J Leszczynski. (1997) Chem.

Phys. Lett. 264, 441. 10 X Wang, L Andrews. (2003) JACS 125, 6581. 11 HJ Himmel, H Schonoeckel. (2002) Chem. Eur. J. 8. 12 P Calvani, C Ciotti, A Cunsolo, S Lupi. (1990) Solid State Communications 75, 189. 13 DC McKean, I Torto, MW Mackenzie, AR Morrisson. (1983) Spectrochimica Acta 39A,

387-98. 14 CJ Bennett, CS Jamieson. (in preparation 2005) Ap. J. 15 AM Coats, DC McKean, D Steele. (1993) Journal of Molecular Structure 320, 269. 16 AD Becke. (1993) J. Chem. Phys. 98, 5648. 17 C Lee, W Yang, RG Parr. (1998) Phys. Rev. B 37, 785. 18 R Krishnan, JS Binkley, R Seeger, JA Pople. (1980) Chem. Phys. 72, 650. 19 D Sillars, C Bennett, RI Kaiser, Y Osamura. (2004) Chem. Phys. Lett. 20 J Cizek. (1969) Adv. Chem. Phys. 14, 35. 21 JA Pople, M Head-Gordon, K Raghavachari. (1987) J. Chem. Phys. 87, 5968. 22 RA Kendall, TH Dunning, RJ Harrison. (1992) J. Chem. Phys. 96, 6796. 23 MJ Frisch, et al. (2002) GAUSSIAN98 Revision A11. 24 RI Kaiser, G Eich, A Gabrysch, K Roessler. (1997) Astrophys. J. 484, 487.

33

MECHANISTIC STUDIES OF TRIFLUOROMETHYL SULFUR PENTAFLUORIDE SF5CF3 :

A GREENHOUSE GASFIRST INFRARED SPECTROSCOPIC

William Carrier Department of Chemistry

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

The formation of SF5CF3(X1A’), through the combination of SF5(X2A1) and CF3(X2A1),

was observed for the first time in low temperature sulfur-carbon matrices at 12 K via infrared spectroscopy upon irradiation of the ices with energetic electrons. The precursor ν1 fundamentals were detected at 857cm-1 and 1110cm-1 for SF5(X2A1) and CF3(X2A1), respectively whereas the trifluoromethyl sulfur pentafluoride molecule SF5CF3(X1A’) was monitored via its absorptions at 846 cm-1, 801cm-1 and 550 cm-1. This formation mechanism, previously hypothesized by Sturges et al. (2000), confirms an alternative source for this potentially dangerous greenhouse gas. These IR data may aid in identifying this gas in prospective atmospheric searches via infrared spectroscopy.

INTRODUCTION

Global climate change is becoming an increasingly important environmental issue. Wide-spread attention is being directed towards the strongest greenhouse gas trifluoromethyl sulfur pentafluoride (SF5CF3) with a radiative force of .59 W m-2ppbv-1 on a per molecule basis.1 Although its concentration of about .12 part per trillion (ppt) as first measured from the Antartic firn in 1999, is relatively small, its concentration is growing at a rate of 6% yr-1.2 Released as a by-product during the manufacturing of fluorochemicals, the only known source, this accounts for only a small existing portion in the atmosphere.3 It was hypothesized by Sturges et al. (2000) that a possible source of SF5CF3 could be the SF5 radical, formed by high-voltage discharge, attacking CF3 groups on the surface of fluorpolymers in high-voltage equipment, indicating this gas could inflict a significant environmental impact in the future. Confirmation of this formation route should then direct environmental assessment in the global use of SF6 and CF3 groups in high voltage equipment. Therefore, these investigations present a valuable role in understanding the formation mechanisms of this gas, in addition to aiding identification in prospective atmospheric searches via infrared spectroscopy.

Although the growth trends of SF6 and SF5CF3 are in agreement, to date, the hypothesized formation route has not been observed from experiments conducted with SF6(X1A1g) and CF4(X1A1).4 We have previously demonstrated, using a surface scattering machine, that in low temperature methane and germane matrices, energetic electrons induce primarily a cleavage of the carbon-hydrogen and germanium-hydrogen bonds to form methyl/germyl radicals and atomic hydrogen.5,6 Here energetic electrons can be used to induce a cleavage of carbon-fluoride and sulfur-fluoride bonds to create SF5 and CF3 radicals and fluoride atoms. Combining an experimental and theoretical approach, we conducted an investigation with low temperature SF6:CF4 matrices, into the formation of the trifluoromethyl sulfur pentafluoride

34

(SF5CF3) molecule; these studies confirm this formation route of the greenhouse gas and identify the most intense infrared absorption frequencies in low temperature matrices. These studies can be expanded to investigate the formation, energetics and also spectroscopic properties of SFxCFy (x= 1- 6; y= 1-6).

EXPERIMENTAL

These experiments were conducted in a contamination-free ultrahigh vacuum (UHV) machine. The main chamber is capable of being evacuated down to 5×10−11 torr by a magnetically suspended turbo pump backed by an oil-free scroll pump. A rotatable highly polished silver mirror contained within the main chamber is cooled via a two stage closed cycle helium refrigerator; connected to a differentially pumped rotary feed through. The silver mirror can be cooled to 10 K and serves as a substrate for the ice condensate. Gas samples can be brought into the chamber through a precision leak valve, which is connected to a gas reservoir and supported by a linear transfer mechanism. To guarantee a reproducible thickness of the ice samples, the deposition system can be moved 5 mm in front of the silver mirror prior to the gas condensation. The SF6-CF4 ices were prepared at 12 K by depositing a pre-mixed gas of sulfur hexafluoride (99.75%) and carbon tetrafluoride (99.99%), at pressures of about 1.4 × 10−7 torr for 30 minutes onto a cooled silver mirror. The spectra of the SF6-CF4 ice appears as a combination of the infrared spectrum of the individual ices previously published.7,8 To determine the thickness of the sample we integrated the infrared absorption features of the ν4 fundamental at 615 [SF6] and 630 [CF4] cm-1. The ice thickness was then calculated using the Lambert-Beer relationship. The integrated absorption coefficients of these fundamentals, 1.45 × 10-17 and 2.08 × 10-18 cm-1, respectively and the density of these ices SF6: 1.4 gcm-3 and CF4: 1.89gcm-3 9,10 determined an optical thickness of 81.5 ± 20 nm [SF6] and 173 ± 20 nm [CF4] providing an estimated total ice thickness of 216 ± 20 nm.

These ices were exposed for 120 minutes by scanning the sample over an area of 3.0 ± 0.4 cm2 with high energy electrons to induce both carbon-fluoride and sulfur-fluoride bond ruptures in the low temperature samples. Irradiation was performed with 5 keV electrons at beam currents of 10 nA and 100 nA at 12K. Background analysis was performed by collecting data immediately prior to the addition of the SF6:CF4 mixture in the UHV chamber. A Fourier transform infrared spectrometer was used for certain identification of the reaction products in the solid state. The Nicolet 6700 DX FTIR unit (5000 – 500 cm-1) operated in an absorption-reflection-absorption mode (reflection angle α = 75º) with a resolution of 2 cm-1. The infrared beam coupled via a mirror flipper outside the spectrometer, was passed through a differentially pumped potassium bromide (KBr) window, and attenuated in the ice sample both prior and after reflection at a polished silver waver. The beam exits the main chamber through a second differentially pumped KBr window before being inspected via a liquid nitrogen cooled detector.

THEORETICAL APPROACH

The molecular structures of various isomers for the SF6, SF5, CF4 and CF3 species were optimized in terms of ab initio density functional B3LYP methods11,12 with the 6-311G(d,p) basis set.13 The coupled cluster CCSD(T) calculations14,15 with the aug-cc-pVTZ basis set16 were also performed at the optimized structures obtained with the B3LYP method in order to compare the relative energies of various isomers. All computations were carried out using the

35

GAUSSIAN 98 program package.17 The relative energies stated in the text are the values obtained with the CCSD(T) method corrected with the zero-point vibrational energies obtained with the B3LYP method. We have performed the vibrational analysis with the second-order Møller-Plesset perturbation theory (MP2 method)18, Hartree-Fock method (HF method), and quadratic configuration interaction method (QCISD method) for several structures in order to examine the dependency of wave functions applied to obtain the vibrational frequencies.

EXPERIMENTAL RESULTS

Infrared absorptions of the trifluoromethyl and sulfur pentafluoride radicals appeared instantaneously with the onset of the irradiation of the SF6:CF4 samples with an electron current of 100 nA at 12K at 1110 cm-1 (ν1; CF3) and 857 cm-1 (ν1; SF5). The position of both these ν1 fundamentals agree soundly with previous experiments.19,20 The strongest mode ν2 of the CF3(2A1) radical is obscured by features of the matrix and could not be undeniably identified. The weaker ν2 fundamental is observable at 664 cm-1. An intensity ratio of the two bands at 1110 and 664 cm-1 is calculated as 1.5 ± 0.5; this is in close agreement with the theoretically computed value. (Table 2). The weakest fundamental ν4 is a factor of 10 lower than the ν2 mode and, hence, is too low to be detected. The strongest mode ν7 of the SF5(2A1) radical is observable at 805 cm-1. Using a scaling factor of 1.09 these data agree nicely with calculated modes. The ν2 fundamental mode is more than a factor of 10 lower than the ν1 mode and also too weak to be detected. The remaining infrared active modes were outside the range of our MCTB detector.

Having identified the precursor molecules, we also detected, immediately after irradiation, the infrared absorption of the trifluoromethyl sulfur pentafluoride [SF5CF3(1A')] at 846, 801, and 550 cm-1. Several of the fundamentals for this molecule overlap, contributing to the observed absorption, and are unobservable individualy. In addition some of these modes have nearly identical intensities and can therefore not be compared with theoretical calculations. The absorption at 801 cm-1 can positively be assigned with a scaling factor of 1.08 as the ν5 fundamental. Although assigned to several modes, the absorption at 550cm-1 is in good agreement with the gas phase mode ν9.21

DISCUSSION AND SUMMARY The infrared data imply that the response of the SF6:CF4 ice upon the electron irradiation is governed initially by sulfur-fluoride and carbon-fluoride bond ruptures which form atomic fluoride plus the SF5 and CF3 radicals. In order to escape either the [SF5…F] or [CF3…F] matrix cage, the fluoride atom needs an excess energy – the lattice binding energy - of a few tens of kJmol-1 to escape. Otherwise, atomic fluoride will recombine with the radicals to react back to the initial molecules. The infrared data suggest that the trifluoromethyl sulfur pentafluoride molecule is formed initially within a single trajectory of the electron. Indicating that neighboring radicals can recombine to the SF5CF3(X1A’) at 12 K if they have the correct recombination geometry. Summarized, the formation of the SF5CF3(X1A’) molecule was observed from the combing of SF5(X2A1) and CF3(X2A1) radicals for the first time in SF6:CF4 matrix at 12 K via infrared spectroscopy.

36

Table 3. Observed species and their absorptions in low temperature SF6:CF4 matrices.

Species Wavenumbers, cm-1 Fundamental SF5 857 ν1

806 ν7

CF3 1110 ν1 664 ν2

SF5CF3 846 ν3,ν4,ν17 801 ν5 550 ν7,ν8,ν9,ν18

820 815 810 805 800 795 790 785 7800.00

0.01

0.02

0.03

0.04

0.05

0.06

Abso

rtpio

n

wavenumbers, cm-1

SF5

SF5CF3

Figure 1: New Absorption feature of the sulfur pentafluoride radicals, SF5(2A1), at 806 cm-1

and the trifluoromethyl sulfur pentafluoride fluoride molecule, SF5CF3(1A'), at 801 cm-1 in the

SF6:CF4 matrix at 12 K

37

865 855 845 835 8250.00

0.01

0.02

0.03

0.04

0.05

Abso

rban

ce

wavenumbers, cm-1

SF5

SF5CF3

Figure 2: New Absorption feature of the sulfur pentafluoride radicals, SF5(2A1), at 857 cm-1 and

the trifluoromethyl sulfur pentafluoride fluoride molecule, SF5CF3(1A'), at 846 cm-1 in the

SF6:CF4 matrix at 12 K

580 570 560 550 540 5300.00

0.03

0.06

0.09

0.12

Abso

rban

ce

wavenumbers, cm-1

SF5CF3

Figure 3: New Absorption feature of the trifluoromethyl sulfur pentafluoride fluoride molecule,

SF5CF3(1A'), at 801 cm-1 in the SF6:CF4 matrix at 12 K

38

ACKNOWLEDGMENTS

These experiments were supported by the NASA Space Grant administered by the University of Hawai’i at Manoa. Special thanks to Dr. Kaiser for providing mentoring throughout the entire research project. The author would also like to thank Corey Jamieson for his assistance in operation of the surface scattering machine.

REFERENCES

1 Nielsen, O. J.; Et al. Atmospheric Environment. 2001, 36, 1237. 2 Sturges, W. T.; Et al. Science. 2000, 289, 611. 3 Santoro, M. A.; Science. 2000, 290, 935. 4 Huang, L.; Zhu, L.; Pan, X.; Zhang, J.; Ouyang, B.; Hou. Atmospheric Environment. 2005, 39,

1641 5 Bennett C.J., Jamieson, C.S. Ap. J. (in preparation) 2005. 6 Carrier, W. J.; Zheng, W.; Osamura, Y.; Kaiser R. I.; Chem Phys. (in press) 2005. 7 Kno1zinger E. Babka, E. Hallamasek, D.; J. Phys. Chem. A 2001, 105, 8176. 8 Forneya, D. Jacox, M. E. J. Chem. Phys. 1994, 101 ,8290 9 Taylor, J.C.; Waugh, A.B. J. Solid State Chem. 1976, 18, 241 10 Smith, J.H.; Pace, E.L. J. Chem. Phys. 1969, 73, 4232 11 Becke, A.D. J. Chem. Phys. 1993, 98, 5648. 12 Lee, C.; Yang, W.; Parr, R.G. Phys. Rev. B 1988, 37, 785. 13 Krishnan, R.; Binkley, J.S.; Seeger, R.; Pople, J.A. J. Chem. Phys. 1980, 72, 650. 14 J. Cizek, Adv. Chem. Phys. 1969, 14, 35. 15 Pople, J.A.; Head-Gordon, M.; Raghavachari, K. J. Chem. Phys. 1987, 87, 5968. 16 Kendall, R.A; Dunning, T.H.; Harrison, R.J. J. Chem. Phys. 1992, 96, 6796. 17 Frisch M. J.; et al. GAUSSIAN98 Revision A11; Gaussian, Inc.: Pittsburgh, P. A., 2002. 18 Møller, C.; Plesset, M.S. Phys. Rev. 1934, 46, 618. 19 Lugez, C.L.; Jacox, M.E., Journal Chem. Phys. 1998, 108, 9639. 20 Forney, D.; Jacox, M.E., J. Chem. Phys. 1994, 101, 8290. 21 Xu, W., Xiao, C., Li, Q., Xie, Y., Schaefer III, H. Molecular Physics. 2004, 102, 1415.

39

DECADAL VARIABILITY IN SLOPE STREAK ACTIVITY ON MARS

David Gremminger Geology and Geophysics

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

The Viking Orbiter Mission of the late 1970’s first captured images of dark, narrow, fan-shaped features extending down sloped regions of Mars. Little was known of these features at the time thus no conclusions could be made as to their nature. The Mars Orbital Camera (MOC) aboard the Mars Global Surveyor (MGS) spacecraft first launched in 1997 also detected these features and revealed that slope streaks are actively forming features on the surface of Mars. Documentation of persisting, faded and new slope streaks was performed in this study to learn more about the nature of these features. Early in the study new slope streaks were confidently identified. New slope steaks are found to be systematically darker than persisting streaks. This discovery raised questions about the fading of slope streaks. If slope streaks are actively forming then they must also fade at about the same rate as formation. Fading is found to occur but the rates of formation and fading are not balanced. Also, fading in some streaks is not uniform and portions of the streaks fade differently than others.

INTRODUCTION

Slope streaks are mass wasting features currently active on the surface of Mars, first studied by Sullivan et al (2001). They can be seen as dark, narrow, fan shaped features extending down sloped dust covered regions of the planet. They show no topographic relief along sharp margins and have no down-slope depositional features. They originate from a point source, are of variable length and width, and display branching down-slope ends (figure 1). Cameras attached to spacecraft orbiting Mars captured the images displaying slope streaks. The first images to reveal slope streaks were obtained during the Viking orbiter mission during 1977 and 1978. Resolution quality in these images is less than 18 meters per pixel, which can make positive identification of slope streaks in certain images difficult. There are no repeat Viking images in which new slope streaks could have been observed. This being said only the most obvious streaks were documented. The most recent images (1997-present) are obtained from the Mars Orbiter Camera (MOC) aboard the Mars Global Surveyor (MGS) spacecraft (Malin and Edgett, 2001). Resolution quality in these images is much more detailed and fosters more convincing documentation of the nature of slope streaks. Observation of the two image sets (Viking and MOC images) is done via personal computer using the Matlab computer software program. Images are downloaded into the program where they can then be adjusted for comparison. Viking images that contain slope streaks are compared to MOC images of the same region and slope streaks are documented. The two overlapping images are compared to determine if there are any new, faded, or persisting slope streaks. All relevant streaks and any other streak activity that may reveal the nature of slope streaks are thoroughly observed and documented.

40

Over the course of this study many new discoveries were made pertaining to the fading, formation and persistence of slope streaks. Bright slope streaks were documented in one overlapping region, are found to persist in much the same manner as dark streaks from Viking to MOC images and no convincing new bright streaks were found.

Figure 1: Ridge showing bright and dark slope streaks form MOC image M2000701. Latitude 10.5oN; Longitude 318.5oW

METHOD OF STUDY

Documentation of images containing slope streaks was performed using the Matlab computer software program. This command-based software has the ability to read raw image data in Planetary Data System (PDS) format and allows for the manipulation of images in terms of rotating, zooming in, zooming out, changing contrast, cropping and isolating areas of images and saving images that have been altered. The method for observing slope streaks is to obtain a Viking image that contains slope streaks, target that region with MOC and then compare the two overlapping images of the same region. To compare the images is to observe them side-by-side using Matlab. Viking images are first looked at to determine any obvious streaks and topographical features in the immediate area that can be associated with less obvious streaks or used as a navigational guide. Once said features are identified, slope streaks in the Viking image are annotated. The MOC image is then studied. If there are any obvious slope streak matches with the Viking image they are documented as persisting slope streaks. Faded streaks are documented by locating a streak in the Viking image and finding the same streak in the MOC image. If the streak is not present in the MOC image the streak is considered to be faded. Next, is to determine if any new slope streaks are found in the overlapping pair by simply documenting streaks that can be seen in the MOC image but are not present in the Viking image.

RESULTS Many new streaks were documented along with the darkness differences between new streaks and persisting streaks. A total of 55 convincing new slope streaks, from eight different Viking/MOC overlapping regions were documented (Table 1). The final number of 55 new slope

41

streaks represents only the most convincing cases of new slope streaks. Ambiguity arises when documenting slope streaks mainly due to the drastic difference in quality between Viking and MOC images. The number of documented convincing new streaks, ~55, is very close to the number of persisting streaks, ~58. Hence, the number of slope streaks appears to have almost doubled in ~24.5 years. This agrees with the formation rate observed in MOC/MOC overlapping images done by Aharonson et al. (2003). We have identified ~55 convincing new streaks but only 11 convincing faded streaks since Viking. Thus it can be said that the rate at which new streaks form is faster than the rate at which streaks fade. Table 1: Regional Counts of New, Persisting, Half-Faded and Faded Slope Streaks. Viking/MOC overlapping images

MOC Lat/Long

Time Difference

New Streaks

Persisting Streaks

Faded Streaks

Half -Faded

f441b09, f441b08,f441b02/ E0501067

27.9oN/ 147.6oW

23.6 11 14 0 0

f441b04 and f441b05/ R1302916

27.8oN/ 147.7oW

26.2 7 11 6 4

f441b04, and f441b05/ R1801921

27.8°N/ 147.7°W

27 0 0 5

0

f748a10/ E0102238 and M2100670

1.8°S/ 344.2°W

22.5 ≥14 ≥3 0 0

f713a70/ R0802542

7.4°N/ 318.7°W

25.4 1 8 0 2

f768a54/ R0901461

10.8°S/ 148.7°W

25.1 ≥15 ≥13 0 0

f748a12/ M0401105

1.7°S/ 344.1°W

21.1 7 4 0 0

*f713a45/ R0901952

9.8°N/ 321.7°W

25.3 ≤4 5 0 0

Total counts Avg. =24.5 ≥55 ≥58 11 6 *Only region containing bright slope streaks New streaks are systematically darker than persisting streaks. Table 2 lists the average darkness values of new and persisting slope streaks from each overlapping pair documented. The scale used is a subjective scale from 1 to 10 with 1 being the least dark and 10 being the most dark. The scale is used to compare the darkness of persisting streaks and new streaks. In every overlap region, new streaks are, on average, always darker than persisting streaks and in most regions, the faintest new streak is still darker than the darkest persisting streak. Because persisting streaks must be older than new streaks, this is direct evidence for fading in all regions studied.

42

Table 2: Average Darkness Values for New and Persisting Streaks. Viking/MOC Overlapping Images Average

New Streaks

Average Persisting Streaks

Minimum New Value

Maximum Persisting Value

f441b09, f441b08,f441b02/ E0501067

8.2 4.5 7 6

f441b04 and f441b05/ R1302916 8.4 3.9 5 6 f441b04, and f441b05/ R1801921 - - 0 0 f748a10/ E0102238 and M2100670

6.9 4.3 5 5

f713a70/ R0802542 10 4.2 10 8 f768a54/ R0901461 7.9 6.3 6 9 f748a12/ M0401105 7.7 4.3 7 5 The most important result of this study was the first ever documentation of fading slope streaks. A total of 11 faded streaks were documented. Due to the fact that this was the first documentation of fading streaks, careful observation of the images containing the streaks was done. Disappearance of slope streaks does not result from resolution or illumination differences. Figure 2 shows the Viking/MOC overlap region where some of the faded slope streaks are located. The Viking image f441b04 is on the left and the corresponding MOC image R1801921 is on the right. The 6 labeled streaks seen in the Viking image (4A-4E and 4O) are clearly not present in the MOC image. The illumination of the sun is more or less coming from the left margin of the page. With the two images having similar illumination directions the streaks seen in Viking should also be seen in MOC but that is not the case. Thus the streaks not visible are completely faded slope streaks. Streaks have disappeared completely in two sets of overlapping Viking/MOC images, while no convincing evidence of fading was found in 6 other overlap regions, including one region within a distance of a few kilometers where many new streaks have appeared but none have faded. This spottiness shows that fading is achieved primarily not by uniform deposition of atmospheric dust, but by localized deposition through either regional atmospheric dust settling or mass wasting events at the surface. Also found are streaks that show unusual fading characteristics. Observation of certain slope streaks reveals that some have been half faded. In total, 6 streaks were confidently identified as half faded streaks. This too is the first documentation of this sort. Figure 3 shows one of the overlapping Viking/MOC regions where half faded streaks are observed. The image on the left (f441b04) is from the Viking orbiter mission in which three easily seen slope streaks are marked with an arrow. They flow downhill which in this image is towards the left margin of the page. The image on the right (R1302916) shows a portion of the MOC image and the three marked streaks are the same as the three in the Viking image on the left. It can be seen that the three streaks are only partially persisting. The MOC image shows that the streaks are half faded. Streaks labeled 4I and 4R show that the top half has been faded while 4H shows the bottom half faded.

43

Figure 2: Viking f441b04/MOC R1801921 overlapping images showing completely faded slope streaks. Latitude 28oN Longitude 148oW

Figure 3: Viking image f441b04 showing streaks 4H, 4I and 4R (left). MOC image R1302916 showing half faded streaks (right). Latitude 28oN Longitude148oW.

44

DISCUSSION Slope streaks are indeed actively forming features, new slope streaks do in fact form and fading does occur. New slope streaks were found in all but one overlapping image pair studied. New slope streaks are not the result of illumination or resolution differences between the two images. If all the new slope streaks documented were present in the Viking images they could have been easily seen because other streaks in the immediate area were seen. Resolution or illumination differences would not cause one streak to be visible and another right next to it not to be visible. After comparing the number of new streaks and faded streaks it can be said that the rate of formation and fading are not in equilibrium. Fading slope streaks identified in this study were documented in two overlapping images. Considerations must be taken when trying to determine if a slope streak is faded or not. The most important is the illumination direction of the sun. If the illumination direction is extremely different from Viking to MOC then it is possible that the streak in MOC cannot be seen. Another is the angle at which the image is taken. If the two images were taken at extremely different angles, streaks may not have the same appearance. After careful observation of the regions that contain the faded streaks, it is clear that illumination or resolution differences do not explain faded streaks and some other mechanism is responsible for the fading. Such mechanisms could be local mass wasting events or atmospheric dust deposition. Global dust storms have occurred in the time between the Viking mission and the MGS mission. Half faded streaks are another area of new discovery presented here. Careful observation of the regions that contained the half faded streaks was done and they too cannot be the result of resolution or illumination differences for the same reasons completely faded streaks cannot result in resolution or illumination differences.

CONCLUSION

This study confirms that slope streaks are indeed actively forming features on the surface of Mars as previously reported by Sullivan et al. (2001). New streaks were confidently identified in all but one of the overlapping regions studied. The number of documented convincing new streaks is almost as large as the number of persisting streaks and the rate at which new streaks form is faster than the rate at which streaks fade. Documented for the first time was the complete fading of slope streaks. Complete fading appears to be regional and may be caused by localized mass wasting events on the surface or localized dust deposition. Streaks showing partial fading in upper portions suggest fading is caused by mass wasting events at the surface and not by global atmospheric dust deposition. Partially faded half streaks also suggest that fading is regionally dependent. One region may be more susceptible to fading than another based on the geographical area in which the streak is located. New streaks seem to be systematically darker than persisting streaks. Gradual fading occurs in most areas as indicated by the darkness values of new streaks compared with persisting streaks. No convincing new bright slope streaks were found.

45

REFERENCES Malin, M.C. and Edgett, K.S. (2001).Mars Global Surveyor Mars Orbiter Camera: Interplanetary cruise through primary mission. J. Geophys. Res. 106(E10), p23429-23570. Sullivan, R., Thomas, P.,Veverka, J., Malin, M., and Edgett, K.S. (2001). Mass Movement Slope Streaks Imaged by the Mars Orbiter Camera. J. Geophys. Res. 106(E10), p23607--23633. O. Aharonson, N. Schorghofer, and M. Gerstell (2003). Slope Streak Form ation and D us Deposition Rates on Mars. J. Geophys. Res. E 108(E12), 5138, doi:10.1029/2003JE002123.

46

SYSTEMS INTEGRATION AND STABILIZATION OF A CUBESAT

Tyson Kikugawa Department of Electrical Engineering

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

A CubeSat is a fully functioning satellite, confined to a box that is ten centimeters on each side and can fit in the palm of your hand. Just as the conventional satellites currently orbiting Earth, the CubeSat is comprised of several subsystems that play a crucial role in carrying out the small satellite’s mission. These subsystems, including communication, power, and data handling must be able to work together as one system in order for the mission to be a success. Systems integration exists to ensure this compatibility among subsystems. In addition, depending on the mission, the CubeSat may need some form of stabilization to keep it from spinning too freely in space. The basic ideas of this have also been investigated.

INTRODUCTION

Satellites play a crucial role in everyday life, from relaying phone calls to protecting the security of an entire nation. These satellites, however, are very large, extremely expensive, and take years to develop. This is where small satellites, such as the CubeSat, have big advantages over their larger counterparts. Small satellites cost much less per unit, so many more can be fabricated for the price of one. Since they are small, they are easier to build and can be produced at a much faster rate. Obviously, one cannot fit all the hardware of a large satellite in the small box that is a CubeSat. Instead, by having many CubeSats, one can construct a network of small satellites that can accomplish the same tasks one large satellite can. Also, should one of these CubeSats in the network some how fail, the network can be reconfigured to pick up the slack. Unfortunately, for the larger satellites, should something go wrong, the entire mission is at risk of failure because of a single fallout. The basic subsystems that make up the internals of a CubeSat are the telemetry, tracking and control (TTC), power generation and distribution (PGD), data command and handling (DCH), and the payload. The contents of the payload depend on the requirements of the mission. Another subsystem that can be included if deemed necessary is attitude determination and control (ADC), which is the stabilizing of the CubeSat and the basics will be discussed later. Each subsystem plays a crucial role in keeping the satellite up and running properly. The main role of the TTC subsystem is the communication link between the satellite and the ground station. The role of PGD is to ensure there is enough power being produced so that all the components of the satellite can operate at the appropriate times. The DCH subsystem is in charge of the overall flow of operations of the satellite, which includes gathering, storing, and processing information from the various subsystems and passing the information on accordingly. Each payload subsystem is specific to the mission that the CubeSat is designed to carry out. The purpose of the payload can vary from temperature tests, image acquisition, or even a biological study. Aside from the mission needs, there are also several constraints that apply to all CubeSats and is discussed next.

47

CUBESAT STANDARDS

The CubeSat standards are limitations and requirements that CubeSat developers are advised to follow when designing and constructing their satellite. The standards were created by the California Polytechnic State University who has also created the launcher that will hold and release the CubeSats once they have reached orbit height. The standards serve as a means to allow the CubeSat launcher to be universal to hold and launch any CubeSat. The launcher is known as the P-POD (Poly Picosatellite Orbit Deployer). The P-POD is basically a rectangular box with a large spring inside. On one face, there is a hatch door where the CubeSat will enter and exit. When a CubeSat is placed in the P-POD, it compresses the spring which, at launch time, will push the CubeSats out into orbit. The P-POD is where the size constraints of the CubeSat is derived from, and the size of one CubeSat is 10 cm x 10 cm x 10cm and weight limit of 1 kg. However, if a single unit size is too small for the parts, a team can opt to increase the size to 1.5, 2, or even 3 units (three is the limit because the P-POD can only hold three single unit CubeSats). By increasing the size of the desired CubeSat, the weight limit also increases. This can be determined by simply taking the single CubeSat limit of 1 kg, and multiplying it by the factor of increase (i.e. if using a 1.5 sized CubeSat, the weight limit is 1.5 kg). As for the housing size increase, only the length is affected. The height and width of the CubeSat remains the same, and like the weight, the length increase is by the same factor of increase in comparison to the single unit CubeSat. So a 1.5 CubeSat, means a 10 cm x 10 cm x 15 cm size limit. Separation springs and switches may also be vital to a mission. The CubeSat, powered by electricity and made of metal, needs a mechanical or electrical means of determining when it has been released into space and is clear to begin its operations. This is where the separation switch would come into play. When the CubeSat is in the P-POD, it will be pressed up against another CubeSat, so by designing the switch to extend slightly out of the structure, it will be pressed down. When the CubeSat has been released, the switch will no longer be registered as being pushed and the satellite can then activate. The springs, on the other hand, help the CubeSat to separate itself from the adjacent CubeSat(s). The suggested placement of the springs is in opposite corners of the point of contact with the other CubeSats. Electrical requirements include a “pull before flight” pin. The purpose of this pin is for those who would like to launch their satellite with their batteries charged. Since the satellites, once inspected, would be sitting on a shelf as they await launch, it would be beneficial to have a mechanism that will disconnect the batteries from the rest of the components so that the batteries do not slowly discharge. The pin would be pulled before they are placed in the P-POD for flight (hence the name), and the separation switches would take over from that point. This brings up another requirement, which is the CubeSat must be powered down while in the P-POD so it will not interfere with the primary payload.

ADDITIONAL SYSTEMS INTEGRATION CONCERNS

Aside from making sure the CubeSat will meet the standards, systems integration also has a few other issues to address. Keeping the size limit in mind, systems integration needs to make sure that all the parts will fit within the confines of the housing. The internal layout is crucial in the success of the CubeSat. Since this needs to be solved prior to ordering the parts, it would be

48

*The camera and Fox will be modified to be lighter (approx 150g total). PCB weight is estimated from a previous project PCB boards.

ideal to have some sort of model of the housing and parts so that a layout can be planned and tinkered with to ensure everything will fit. As mentioned before, systems integration needs to ensure all the subsystems are able to come together as one system. They needs a means of connecting with each other and systems integration will need to determine how these interconnections will be made. This is done when the layout and printed circuit boards are being designed. These connectors are very important since they are the means of which the subsystems will be communicating with one another, and a poor selection could result in a system failure down the road. Once parts have been received, each subsystem should be testing the parts specific to their subsystem separately, keeping in mind what is needed of them so that their parts will be able to cooperate with the other subsystems. After each part and subsystem has been developed and tested, the integration process begins. A suggested method would be to test only a few pairings/groupings of subsystems at a time, then start to put it all together and finish off with a complete system test. This final test procedure could go as far as simulating the actual flight process as it goes into its orbit.

APPLICATION

The University of Hawaii CubeSat team is jointly working on a CubeSat with an electrical engineer from a local engineering firm. The primary mission of this CubeSat is image acquisition for disaster mitigation purposes. Therefore, the primary payload will be an on board camera. Also included in the payload, is a global positioning system (GPS) and a nano-inertial measurement unit (nIMU). These two devices will be used for geo-referencing the images taken so the location of the image can be readily determined. The engineering firm has been placed in

charge of developing this system, while the duty for UH is to develop the system bus (TTC, PGD, and DCH subsystems, and the structure). This iteration will be built as an engineering model, so certain aspects that may be important for a flight ready system, will be left out. Due to the large number of parts, a one and a half sized CubeSat will be used. The first matter (and easiest) of business is the weight budget. Since a 1.5 unit cube can be allotted 1.5 kg of total weight, it would be good to make sure it will not exceed that weight or it could devastating (or it would just require selecting different, smaller parts). Table 1 shows the current weight situation the UH CubeSat. The other major constraint set by the 1.5 unit cube, is the 10 cm x 10 cm x 15 cm size limit. Figure 1 shows the layout of the parts being used within the CubeSat housing (drawn by UH mechanical engineering graduate student, Lance Yoneshige). There is sufficient room for the connectors and wire (not shown) which will be needed to connect each of the circuit boards. There will be five separate circuit boards in the

Table 1. Weight Budget Part Weight (in

grams) MHX2400 (radio) 68.33 Structure 398.58 Battery pack 91.11 GPS Antenna 22.67 GPS Receiver 11.33 PCB boards ~138.95 Fox board ~90.72 Camera ~209.78 nIMU 11.33 Solar cells ~47 Miscellaneous ~400 Total 1489.8

49

Figure 1. Internal Layout – Top View. (Courtesy Lance Yoneshige) -1- Axis camera -2- nIMU board -3- Fox board -4- GPS antenna -5- GPS receiver -6- battery box -7- PGD board -8- TTC/PDG board

CubeSat. Two full boards, one for the PGD subsystem circuits, and the other that will also have PGD components, but it will be shared with the TTC part, the MHX2400 radio from Microhard. On the other side of the satellite, there will be two half boards, one for the nIMU and the other

for the Fox board (the central processor). Finally, there will be a mini board, that will contain just the GPS receiver and it sits perpendicularly on the PGD board. As far as the structure is concerned, it will contain a separation switch, but due to cost concerns, the springs will be left out (as mentioned earlier, the project is only an engineering model). Fortunately, the original structure design had springs, so if needed, a quick adjustment can be made to include the springs. It has been manufactured by 3V, a company in Tacoma, Washington, and except for a few adjustments (due to late design changes), it is in good shape and up to specifications. The project has not reached the level of any system testing thus far due to unforeseen setbacks, but the individual subsystem development has been moving fairly well and it should not be too large of a task to integrate the subsystems together.

STABILIZATION

Stabilization is a daunting task for any small satellite. Although, the design and development of any stabilization system is quite complicated no matter the size of the satellite, it becomes a greater problem for a CubeSat because of it size and weight limitations. For larger satellites, it might be easier to design because it has a lot of room and a much greater weight budget to put bigger, better measurement equipment. By having more of these devices, it could make it slightly easier to implement such a system. The CubeSat on the other hand, does not have this luxury. It can only house a limited amount of stabilization equipment while still requiring room for the rest of the subsystems. As the name suggests, the purpose of the stabilization subsystem, or attitude determination and control (ADC), is to be able to determine the orientation of the satellite in space and be able to redirect the way the satellite is facing to a more desirable direction. This application could be useful in a CubeSat such as the one described above, but due to the difficulty of the design and the fact it is simply an engineering model, there will be no ADC included in it.

DISTURBANCES IN SPACE

50

While in orbit, the CubeSat will be subject to certain environmental torques that will cause it to spin freely. Environmental torques are external forces that could cause the satellite to rotate in certain directions. The first of these torques is the gravitational field. The force of gravity, though much smaller in space, still has an effect on the satellite. Gravity exerts its greatest force on an object on the object’s center of mass. If the center of mass is not in the geometric center of the satellite, gravity will favor the side that the center of mass of the satellite is closer to, and it will cause the satellite to rotate in that direction. The Earth’s magnetic field is also an important one. Even without magnets in the satellite, the magnetic field can still have an effect on the satellite. The residual magnetic field of the satellite that is produced by the magnetic moment of the satellite, eddy currents, and the hysteresis effect on the materials of the satellite. Another possible disturbance is solar radiation. This comes from the electromagnetic radiation from the sun. The effect of this disturbance is related to the orbit height, and the geometry of the surface that is being hit. A disturbance that will normally affect low orbits, is a sort of aerodynamic torque. This torque is a result of the molecules in the upper atmosphere colliding with the sides of the satellite, and like the effect of solar radiation, the effect of this torque relies on the geometry of the satellite faces being hit. Sometimes certain devices in the satellite can produce disturbances to the satellite itself and needs to be considered (depending on the device, this could be desirable and is used to actually control the satellite instead of cause it to venture off target). An example of an undesirable torque produced by the satellite upon itself is the sloshing of a liquid. This case is usually for micro-thrusters, where a liquid is sometimes used a thrusting fuel and as it sloshes around in its container, could jolt the satellite.

PASSIVE STABILIZATION

The idea of stabilization is to take advantage of these torques to steer the satellite in a desired direction. There are two types of stabilization; passive and active. Passive stabilization is a method which interacts with the environmental torques to reorient itself. There are two primary ways to implement this, and that is to use the gravitational field and/or magnetic field of the Earth. The simplest way to use the magnetic field is to include magnets in the satellite and setting them up in such way that it will line up with Earth’s “magnet”. Hysteresis rods may also be included to damp the rotation of the satellite and get the satellite to become more and more stabilized. Using the gravitational field is also a fairly simple idea (again, the idea is simple, but the implementation and design could get very messy), and can be done by offsetting the center of mass so that gravity will pull the side that the center of mass is closest to, down toward Earth. This can be done by extending a gradient boom, which is simply a mass extended out to shift the center of gravity towards that mass.

ACTIVE STABLIZATION

Active stabilization is the type of stabilization that requires the satellite to actively take part in changing its own orientation. This is done by utilizing data taken from on-boards sensors and then using some type of torque producing mechanism to adjust the satellite to an orientation that is practical to the mission. These on-board sensors include magnetometers that measure the

51

direction of the magnetic field relative to the satellite; accelerometers that measure the direction of acceleration of the satellite; sun sensors, that determine the location of the sun relative to the satellite; and gyroscope that measure the rotation of the satellite and the direction it is facing relative to an origin. In terms of torquers used, some common selections are micro-thrusters, that use liquid or gas, and shoot out a small amount to force rotation in the opposite direction of the thrust; electromagnets, which is simply a coil of wire that has a current passed through it (this follows the phenomenon that a current flowing through a wire coil acts as a magnet); and momentum wheels that spin within the satellite and use rotational motion effects to redirect the rotation of the satellite.

DISCUSSION/CONCLUSION

Systems integration, though not a subsystem, playas much more crucial role in the later development of the satellite than in the beginning. If the group working on the project is smaller, it is much easier to coordinate all the parts and interconnections between the individual subsystems. However, as the CubeSat group size increases, it becomes much more difficult to coordinate, in which the role of systems integration greatly increases, since they will need to make sure each subsystem is getting what they need from the other systems. Passive and active stabilization each have their pros and cons, and it is up to the developer to select the method that is best for their mission. But some key points should be noted, although active gives the satellite more control over itself, it does require a much greater dedication to the development of this type of system. It requires much more code and design for such a system to be successful. Also, active stabilization would put more pressure on the PGD subsystem to produce more power because of the added requirements of the sensors and activating the torquers. Of course, if the team has the resources and manpower to do so, it may be an advantage to have an active system.

ACKNOWLEDGEMENTS

I would like to thank my mentor, Dr. Wayne Shiroma, for his support; the Hawai’i Space Grant Consortium, for funding my project and their aid in our trips to Colorado and Utah for small satellite workshops; and the entire University of Hawaii CubeSat team.

REFERENCES “CubeSat Documents – For CubeSat Developers”. California Polytechnic State University

CubeSat website. <http://cubesat.calpoly.edu/_new/> CubeSat Workshop. 8-11 Aug. 2005. Utah State University. Murakami, B. (2005) The Design, Fabrication and Testing of a Nanosatellite: A How To

Approach. Student Paper, University of Hawaii-Manoa. Wertz, James R., Spacecraft Attitude Determination and Control, 1978.

52

IMPROVING THE DECONVOLUTION METHOD FOR ASTEROID IMAGES: OBSERVING 511 DAVIDA, 52 EUROPA, AND 12 VICTORIA

Z Robert Knight

Department of Physics and Astronomy University of Hawai`i at Hilo

ABSTRACT

Deconvolution of astronomical images is a process whereby images are modified in an

attempt to remove the blurring effects of our turbulent atmosphere and the inherent systematic errors that result from use of an Adaptive Optics (AO) system. We have used an image deconvolution program called MISTRAL1 to deconvolve AO images of three asteroids taken with Keck II’s NIRC2 camera: 511 Davida, 52 Europa, and 12 Victoria. We were looking for geologic surface features on the asteroids, as well as asteroid size and pole position for rotation. By understanding more of the structure and composition of these asteroids, we can learn more about how the solar system formed and evolved. Along with having used MISTRAL on these asteroid images, we also tested and evaluated the use of the software, in order to learn to use it to its best ability. Here we describe an effective way to use MISTRAL. An improved method of deconvolution will be beneficial to all astronomical AO imagery, which could improve scientific astronomical measurements.

INTRODUCTION

No telescope image is a perfect representation of an astronomical object, even with the

best Adaptive Optics and cameras. Images are affected by everything in between the object emitting light and the computer screens on which they are eventually displayed, i.e., images are spatially convolved and noisy. MISTRAL (for Myopic Iterative Step-preserving Restoration Algorithm) is an image restoration method written in Interactive Data Language (IDL) which can deconvolve images, based on prior assumptions about the objects being viewed, a noise and convolution model, and a point-spread-function (PSF).

We have used MISTRAL to deconvolve images of three large, main belt asteroids, 511 Davida, 52 Europa, and 12 Victoria, as part of the Resolved Asteroid Project (RAP). These images were taken with the Keck II telescope with the Adaptive Optics assisted NIRC2 camera. 511 Davida was observed December 27, 2002, 52 Europa was observed January 20, 2005, and 12 Victoria was observed June 11, 2003.

Previously, these asteroids have only been seen as unresolved point sources. Some properties of the asteroids, such as mass, period of rotation, and rotation vector have been measured or estimated with photometric light curves. Now, with the use of a large telescope, AO, and deconvolution, we can more precisely determine the sizes, rotation vectors, and albedos of these asteroids, and have begun to create improved 3D models of them. With these improved models, we also hope to enable ourselves and other researchers to better constrain the physical geology and impact history of asteroids, which when understood, will provide clues to understanding the processes that govern planet formation.

1 MISTRAL: COPYRIGHT (C) Conan Mugnier Fusco - ONERA 1998-2000.

53

To use MISTRAL, one can simply supply it with an image, PSF, and two numeric parameters, but for the best results, specific pre-deconvolution processing should be applied to the image and PSF, as well as a careful selection of the parameter settings. We have explored several ways to prepare images for MISTRAL, and recommend what worked best for us.

DECONVOLUTION

The deconvolution process attempts to un-blur images by modeling the way in which

light from the object being imaged spreads out. To model this spread, the deconvolution routine MISTRAL uses an image of a point source (star) nearby the object to act as a point-spread-function (PSF), which is a measurement of the spread of light in an image (Figure 1). This is classical mode. Alternatively, MISTRAL can work without a known PSF in myopic mode. The MISTRAL deconvolution routine has been validated by modeling and use on some planetary objects (Mugnier, 2004).

Figure 1: These images show (a) asteroid 52 Europa, reduced, (b) a PSF for deconvolution, and (c) the same image of 52 Europa, deconvolved.

The three main components of the deconvolution method are a fine noise model, a PSF

estimation capability, and an object regularization term. The object regularization term (object prior) is a way to include prior knowledge of the observed object into the deconvolution method. MISTRAL uses a prior that is suited to astronomical objects that have a mix of sharp edges and smooth areas. (For pointlike objects, an alternative prior can be used.)

The convolution model used by MISTRAL is: i = (h*o)+n, where i is the image, o is the object, h is the PSF, * is the convolution operator, and +n is the (not necessarily additive) noise. To use MISTRAL on a raw image i, it must be preprocessed, which should involve the correction of the background and flat field, the camera’s bad pixels and correlated noise, the scaling of the image in photons, and the recentering and addition of images. The PSF h is usually found by recording the corrected image of a nearby, unresolved star.

In classical mode, deconvolution involves a noise model and an object prior, and works by finding the minimization of a criterion, which MISTRAL does by following a Bayesian probabilistic or penalized-likelihood method. The noise model is designed to account for photon noise, following Poisson statistics, as well as detector noise, following Gaussian statistics. The object prior that is used is quadratic-linear, or L2-L1, which means that it tries to use the best

54

aspects of both quadratic and linear priors. This prior has two hyper-parameters; the global factor μ (regobj) and the threshold δ (thresh). The global factor μ determines whether the object prior tends toward linear or quadratic regression, while the threshold δ determines the threshold in the model at which deconvolution will tend to switch between quadratic and linear regression. In MISTRAL, these must be set as parameters in the call to the MISTRAL function. A recommended set of hyper-parameters is to take μ as about 1, and δ to be on the order of the image gradient’s norm (Mugnier, 2004).

In myopic mode, the PSF and the object are jointly estimated in the same probabilistic framework. This mode should be used when the observed PSF is not available. The minimization of the probabilistic criterion is started with an estimate of the object for a fixed PSF, and stopped when the object and PSF no longer evolve.

PRE-DECONVOLUTION

Prior to deconvolution, the NIRC2 asteroid images had to be reduced as usual, which

included flat field division, background sky subtraction, removing hot/dead pixels, cropping the image to the asteroid, and coaddition of frames.

Since there were no images taken for use exclusively as sky frames, sky frames had to be created from the asteroid images. There were two methods tested for this: one of them involves median filtering and the other involves averaging over quadrants of the images which do not have asteroids in them. The only difference in the resulting deconvolutions between these methods that was apparent was the number of iteration times that MISTRAL used before it decided upon convergence. It was decided that either sky frame creation method works equally well.

On the other hand, MISTRAL is very sensitive to coaddition. Image coaddition is a process whereby successive images of the same object or PSF are aligned with one another and added together, in order to produce an image which has a higher signal to noise (s/n) ratio than the individual images had. Statistical calculation tells us that when N images are added together or averaged, the resulting s/n ratio increases by sqrt(N). Since higher s/n ratios indicate less noisy images, image and PSF coaddition is an important part of the pre-deconvolution data reduction process.

Figure 2: The image on the left is a reduced and (improperly) coadded image of 52 Europa, and the image in the center is its deconvolution. Notice that even though the coaddition

55

looks good to the eye, MISTRAL shows us that the image was not well coadded. The right image is the deconvolution after the coaddition was fixed.

There were 3 coaddition methods investigated. The first one, aligning images by the

brightest pixel within a region, turned out to be problematic for both image and PSF addition, producing blurry images and PSFs. This is evident in our early attempts at deconvolution, which showed artificial checkerboard or striped patterns (Figure 2). The second method, aligning images by use of the IDL Astronomy Use Library’s image correlation routine, correl_optimize, appears to be effective for images but not for PSFs. The third method involves the user clicking on the center of the PSF then calling IDL’s gauss2dfit routine and aligning the Gaussian fits. Coadding the images with correl_optimize and the PSFs with gauss2dfit produces the best deconvolved images.

SETTING THE PARAMETERS

To deconvolve images, MISTRAL requires that the user supply two numeric parameters: regobj and thresh. It is not obvious what numbers should be used for these in order to come up with the best results. One method recommended by other MISTRAL users is to simply try various parameter settings in the program and see which ones give the best results. In order to determine the best parameter settings, we developed an IDL procedure to deconvolve an image with a PSF over a range of parameter settings, and give back an HTML table of the resulting images. The best parameter settings can now be selected from the table.

As of yet, there is no definitive quantitative method to determine which deconvolution is the best. Deconvolved images have so far been compared by how many iterations MISTRAL needs to deconvolve an image, and by qualitative evaluation of the sharpness of the asteroid image edges and the amount of ringing visible within the asteroid images. One way that an image might be better than another could be a less prominent ringing effect in the images. Another indicator of a better image would be the appearance of asteroid surface features in the deconvolved images.

In deconvolving images of these 3 asteroids, it was noticed that the parameters had to be set differently for each asteroid. In the case of 12 Victoria, which had the highest change in luminosity between frames, different parameter settings were used on different frames.

ASTEROID RESULTS

The immediate benefit to using MISTRAL on asteroid images is that the edge of the

asteroid becomes very apparent, and so the asteroid’s size and shape can be immediately determined. This size measurement puts upper limits on the sizes of the asteroids. From these improved size measurements, the albedos of the asteroids can be calculated.

Prior to deconvolution, we had coadded images of 511 Davida at 11 timesteps spanning its full 5.13h period with 7 PSFs, 7 timesteps of 52 Europa spanning most of its 5.63h period with 2 PSFs, and 6 timesteps of 12 Victoria spanning 5.55h of its 8.66h period with 1 PSF. Images of 511 Davida and 12 Victoria were all taken in the near infrared K’ filter, while images of 52 Europa are in each of the K’ and H filters. Table 1 gives the UTC and phase angle of each timestep for Europa and Victoria.

56

Using IDL three-dimensional object graphics, Conrad et. al. showed using a rotating triaxial ellipsoid that the size of 511 Davida is near the lower bounds of the size estimate given by Drummond and Hege. Using a similar technique, we compared the relative sizes and rotation vectors given by Michalowski et. al. for 52 Europa and those given by Torppa et. al. for 12 Victoria, to the apparent sizes and rotations seen in our resolved, deconvolved images.

To analyze the shapes, sizes, and rotation vectors of 52 Europa and 12 Victoria, we first found their edges in the deconvolved images using a luminosity threshold of 1/3*max. We then used IDL to plot 3-D ellipsoids at the pole orientations and shapes predicted for the asteroids. For 52 Europa, the axial ratios are given as a/b = 1.21 and b/c = 1.04. There are two possible rotation vectors given in ecliptical longitude and latitude as 262°, +46°, and 67°, +25° (Michalowski, 2004). For 12

12 Victoria period: 8:40 52 Europa period: 5:38 Phase UT UTREL θREL Δθ Phase UT UTREL θREL Δθ 1 6:20 0.00 1 10:40 0.00 2 7:34 1:14 51.23 51.23 2 11:27 0:47 50.06 50.06 3 9:25 3:05 128.08 76.85 3 12:03 1:23 88.40 38.34 4 10:18 3:58 164.77 36.69 4 13:03 2:23 152.31 63.91 5 11:53 5:33 230.54 65.77 5 13:46 3:06 198.11 45.80 6 12:45 6:25 266.54 36.00 6 14:17 3:37 231.12 33.02 7 15:04 4:24 281.18 50.06

Table 1: Times and rotation angles for the epochs of the 52 Europa and 12 Victoria

observations. UTREL gives the elapsed time from the first epoch; θREL gives the rotation angle since the first epoch in degrees; and Δθ gives the rotation angle between epochs in degrees.

Victoria, the axial ratios are given as a/b = 1.3 and b/c = 1.3. The rotation vector is 137°,

+55° (Torppa, 1993). The edge outlines were then laid over the ellipsoids, and the predicted shapes were compared with the observed ones (Table 2).

As can be seen in Table 2, neither 12 Victoria nor 52 Europa are as ellipsoidal as 511 Davida appeared to be (Conrad, 2006). For these two asteroids, the elliptical triaxial ratios seem to be nearly correct, but the rotation axes seem to be wrong.

12 Victoria

52 Europa Pole 1

52 Europa Pole 2

57

Table 2: The model ellipsoids with the deconvolved asteroid image outlines.

MISTRAL PERFORMANCE RESULTS

MISTRAL has been tested with various parameter settings on images of 511 Davida, 52 Europa, and 12 Victoria. In the pre-deconvolution data reduction, there were two methods of creating sky frames tested, and both worked equally well. It seems that deconvolution is not dependant on how sky frames are made. The coaddition method used to stack images is important in deconvolution. Of the 3 coaddition methods tested, image correlation worked best for asteroid images, while elliptical Gaussian fit correlation worked best for PSFs.

The MISTRAL parameters regobj and thresh can be selected by choosing them from a table of deconvolved images. The luminosity of the image to be deconvolved seems to have a strong affect on the best choice of parameters. The method for choosing the best deconvolution is qualitative: the best images have sharp edges, little ringing, and possibly surface features.

Half of the images of 52 Europa were taken with the K’ filter and the rest were with the H filter. Images taken in the H band are noticeably noisier than those taken in the K’ band. This is apparent in the reduced data as well as in the deconvolved images, which showed the size of the asteroid as slightly larger in the H color than in K’.

CONCLUSION

With the successful use of deconvolution, the edges of the asteroids became sharp in the

resolved images. This enabled a comparison to be done between the apparent edges of the asteroids and the triaxial ratios and rotation models of the asteroids. The next steps in analyzing these asteroids will be measuring their sizes and albedos, and constructing better rotation models using ellipsoids. This modeling will be much more effective with many more frames over the course of one asteroid rotation. Six or seven frames gives only a small snapshot compared to the nearly continuous photometric measurements that have been done. With more frames available per cycle, shape and rotation modeling will be much more effective and accurate.

ACKNOWLEDGMENTS

Thanks to Al Conrad, the Resolved Asteroid Project (RAP), and W.M. Keck Observatory

for the data used in this project. Thanks also to Christophe Dumas, for providing MISTRAL for use on the RAP.

58

REFRENCES Conrad, A., et. al. (2006) Rotation and morphology of asteroid 511 Davida. Lunar and

Planetary Science XXXVII. Drummond, J. D., and Hege, E. K. (1989) Speckle Interferometry of Asteroids. Asteroids II,

171-191.

Michalowski, T., et. al. (2004) Photometry and models of selected main belt asteroids I. 52 Europa, 115 Thyra, and 382 Dodona. Astronomy & Astrophysics. 416, 353–366

Torppa, J., et. al. (2003) Shapes and rotational properties of thirty asteroids from photometric

data. Icarus 164 346–383 Mugnier, L.M., Fusco, T., and Conan, J.-M. (2004) MISTRAL: a myopic edge-preserving

image restoration method, with application to astronomical adaptive-optics-corrected long-exposure images. J. Opt. Soc. Am.

63

A THEORETICAL INVESTIGATION ON THE KINETIC AND RADIATIVE EXTINCTION OF SPHERICAL DIFFUSION FLAMES IN MICROGRAVITY

Kin Wai Leung

Department of Mechanical Engineering University of Hawai’i at Manoa

Honolulu, HI 96822

ABSTRACT

Thrust and fire safety are among NASA’s major concerns in the fulfillment of its mission on Mars Exploration, especially the missions that planned to bring astronauts to and back from the Red Planet. Responding to these concerns, the research studied the burning characteristics and extinction of diffusion flames in space, which are different from those on Earth due to the lack of gravity. Because microgravity combustion experiments require special facilities available only in NASA and are very costly, the investigation is theoretical that includes analytical and computational contents. In the analytical study, a mathematical model was formulated to identify the flame location, flame temperature, and extinction condition. The computational study was performed using an existing flame code that incorporated detailed chemistry and transport properties. The analytical study focuses on the steady burning while the computational study on the transient behavior of flames. Spherical flames stabilized by a porous burner were used in this research. It was observed that Lewis number of fuel in the ambient has a profound impact on the kinetic extinction state. From the numerical analysis, it was observed flames with a higher flow rate extinguish quicker due to the radiative heat loss .

INTRODUCTION

In recent years, the primary interest of NASA is focused on the exploration of Mars, which includes launch of spacecraft carrying astronauts to and back from Mars. Thrust and fire safety are among NASA’s major concerns to fulfill this most challenging mission. In a space travel, as on earth, thrust is provided by the reaction between the fuel and the oxidizer. To provide thrust, the flame is required to burn as strong as possible and extinction should be avoided. During the trip, fire outbreak can also occur from various ignition sources such as radiation from the Sun. Such accidents need to be prevented because external assistance is unavailable when a fire burst out and the result could be disastrous. If a fire accident happens, the flame needs to be extinguished as fast as possible to minimize the damage. In most of practical combustion systems, including those used to provide thrust in space travels and related to undesired fire incidents, the flames are non-premixed. The fuel and oxidizer are supplied from separated sources and transported through convection and molecular diffusion to the reaction region within which they meet and react to produce products and heat. The flame is also known as diffusion flame.

Realizing the needs, the research studied the burning characteristics and extinction conditions of diffusion flames stabilized by a spherical porous burner in microgravity. By employing this type of burner, different factors that affect the flame can be separated in order to study the fundamental behavior of diffusion flame extinction. With this geometry, reactant is supplied uniformly from the burner into the ambient filled with other reactant. After the ignition,

64

Inert Region

Burner Exit

Inert Region

Branching Region

Combination Region

Figure 2: Flow domain split into five regions

a spherical flame will propagate out until it is extinguished, steady state is reached, or it establishes a position where fuel and oxygen are totally consumed. The flame is spherical due to lack of gravity in space. One of the advantages of this geometry is being able to control the convection direction from fuel to oxidizer (if fuel in supplied from the burner) or from oxidizer to fuel (if oxygen is supplied from burner). Also, this geometry allows for the control how much the inert gas (primarily

nitrogen) will be supplied in the fuel side and how much in the oxygen side. Due to this geometry, four limiting cases arise: 1. fuel supplied from the burner into an

ambient with air; 2. diluted fuel supplied into oxygen; 3. air supplied into fuel; and 4. oxygen supplied into diluted fuel. Existing studies show that Lewis number of the ambient reactant have a strong effect on the flame properties or flame characteristics. Lewis number is the ratio of characteristic mass diffusion and characteristic thermal diffusion. For the first two cases, where oxygen is in the ambient, studies show that Lewis number is close to unity, not giving much insight on how it affects flame properties. When fuel is in ambient, however, Lewis number can be away from one. Therefore, in order to better understand the effect of Lewis number on flame properties, problem where the ambient is filled with fuel, namely ethylene, was studied for the analytical research.

The research studied the above cases with two approaches. The first approach was an analytical investigation which consisted of formulating a set of equations based on the natural laws of mass, species and energy conservations, and then solving the resulting equations by activation energy asymptotics to identify the flame location, flame temperature, and extinction condition. The second approach was a computational study, using a flame code already developed to study the transient behavior of C2H4/O2/N2 flames under various given conditions.

Analytical Method In the analytical investigation, a mathematical formulation that includes the conservation of mass, species concentration, and energy was formulated to identify the flame location, flame temperature, and extinction condition. In the study, the flow domain was split to five subdomains; a very thin branching region in which the branching reaction occurs, two broader but still thin

Porous Region

Core Region Inner Gas Region

Flame Sheet

Figure 1: Schematic of a spherical diffusion flame

Outer Gas Region

r

br

65

oxidation regions that sandwich the branching region, and two broad chemically inert transport regions out of the termination regions. Each of these three groups of regions was governed by different transport processes. Equations in each of these regions were solved individually and unified through the matching between two neighboring regions. Radiation was not included so only the kinetic extinction, meaning the limit that reaction rates becoming too slow to support steady burning, was studied.

In the limit of infinity fast reaction or called reaction-sheet limit, all of the reactants are completely consumed, such that the adiabatic flame temperature given by

˜ T f ,0 = ( ˜ T s + ˜ Y O ,s ) − ( ˜ T s + ˜ Y O ,s − ˜ T ∞ )[ ˜ Y O ,s / (1 + ˜ Y O ,s )]1/ LeF

(1)

is obtained, and the location of the flame sheet is

˜ r f = LeF ˜ m / n (1+ ˜ Y O,s

−1 ) (2)

In the above, sT~ the supplied temperature of the gas at the center of the burner, ∞T~ the ambient

temperature, sOY ,~ mass fraction of the oxidizer supplied from the burner, FLe is the Lewis

number of fuel, and m~ is the mass flow rate. The ~ represent nondimensional quantities defined as such

∞

=,

,~FF

rgpr Yq

TcT , ˜ Y O =

νF WFνO WO

YOYF,∞

, ˜ r = r

rb,

˜ m =

cp,g m4π rbλg

=4π r2cp,gρgu

4π rbλg=

rb2˜ r 2cp,gρgu

rbλ g= (ρgu)˜ r 2

rbcp,gλg

Lei =

λg / cp,gρg Di

, ˜ ρ g = ρg/ ρg ,f

where r the spatial coordinate along the radial direction (See Figure 1), br the outer radius of the porous burner (See Figure 1), rT the temperature at location r, gpc , the specific heat of the gas at constant pressure, ∞,FY the mass fraction of the fuel at the ambient, OY the mass fraction of the oxidizer, Fq heat of combustion per unit mass of the fuel, Fν the stoichiometric coefficient of the fuel, Oν stoichiometric coefficient of the oxidizer,

FW the molecular weight of the fuel, OW the molecular weight of the oxidizer, m mass flow rate, u radial flow velocity along the radial direction, gρ the gas density, fg ,ρ gas density at the flame sheet, gλ the thermal conductivity of the gas, iLe the Lewis number of species i , iD mass diffusion coefficient of species i

To account for finite rate reactions with high activation energies, a simplified two-step reaction mechanism given by

66

νF F + R → 2 R (3)

R + νO O → P (4)

was adopted, where F is the fuel, R the radical, and P the products. Both reactions are of the high activation energy kind. The branching reaction, (3), represents the pyrolysis of fuel molecules for the production of radicals, is a very high activation energy and slightly endothermic reaction. The termination reaction, (4), symbolizing the combination of radicals to produce final combustion products, has relatively lower activation energy but is highly exothermic. Only the chain branching and termination reactions were considered since these two types of reactions will lead to extinction.

Since reaction rates are not infinity fast, a small amount of reactant leakage across the flame and a small flame temperature reduction occur:

The analysis yielded a structure equation

d2 Θ1± / d ζ2 = Λ 2 Φ R,1

± ΦO,1± exp ( − Θ1

± ) (5)

with boundary conditions

Θ1−(ζ → −∞) = − ˜ T b,1exp [ ˜ m (1− ˜ r f−1)] −( ˜ T f ,0 − ˜ T s)( ˜ m / ˜ r f2 )ζ (6)

(dΘ1− / dζ )ζ → −∞ = −( ˜ T f ,0 − ˜ T s )( ˜ m / ˜ r f2) (7)

Θ1+(ζ → ∞) = −aT ,1

+ [1− exp (− ˜ m / ˜ r f )]+ [ ˜ Y O,s− ( ˜ T f ,0 − ˜ T s )]( ˜ m / ˜ r f 2)ζ (8)

(dΘ1+ / dζ )ζ → ∞ = [ ˜ Y O,s−( ˜ T f ,0 − ˜ T s)]( ˜ m / ˜ r f2) (9)

Θ1−(ζ = 0) = Θ1

+(ζ = 0) = ˜ T f ,1 (10)

( d Θ1− / d ζ )ζ =0 = ( d Θ1

+ / d ζ )ζ =0 (11)

where 1Θ the temperature reduction from the adiabatic flame temperature in the reaction region,

1,RΦ is the radical concentration in reaction region, 1,OΦ the oxidizer concentration in reaction region, ζ the stretched coordinate in the reaction region, 2Λ the reduced hleroDamk number representing the ratio of the characteristic flow time and the characteristic reaction time, 1,bT the deviation of burner exit temperature from the reaction-sheet limit, and 1,Ta an integration constant representing the temperature decrease from the reaction-sheet limit in the chemically inert region. The expressions of 1,R

±Φ and 1,O±Φ were solved from the analysis as

ΦO,1± = LeOΘ1

± + aT ,1+ [1− exp (− ˜ m / ˜ r f )]− ( ˜ T s+ ˜ Y O ,s− ˜ T f ,0)( ˜ m / ˜ r f 2)ζ − aT,1

+ [1− exp ( − LeO ˜ m / ˜ r f )] (12)

67

ΦR,1− = LeR Θ1

− + aT ,1+ [1− exp (− ˜ m / ˜ r f )]+ ( ˜ T f ,0 − ˜ T s )( ˜ m / ˜ r f2)ζ − aT ,1

+ [1− exp (− LeR ˜ m / ˜ r f )] (13)

ΦR,1+ = LeR Θ1

+ + aT ,1+ [1− exp (− ˜ m / ˜ r f )]− ( ˜ T s+ ˜ Y O ,s− ˜ T f ,0)( ˜ m / ˜ r f 2)ζ − aT,1

+ [1− exp (− LeR ˜ m / ˜ r f )] (14) where OLe and RLe is the Lewis numbers of the oxidizer and reactant, respectively. Equation (5) subjected to the boundary conditions (6)-(11) needs to be integrated numerically. A code using a fourth order Runge-Kutta Method was developed and the solutions of the T1 (0), representing reduction of the flame temperature from the adiabatic limit, versus the Lamba2Bar ( 2λ , rescaled reduced hleroDamk number) for selected Lewis numbers were obtained. Varying the Lewis number of a reactant while fixing the other’s, the reactant’s effects on the flame properties were studied. For example, varying fuel’s Lewis number while fixing the Lewis numbers of radical and oxidizer, fuel’s effects on flame properties were only observed. Whereas, the analytical analysis examined the steady burning of diffusion flame in microgravity, the numerical method will study the transient behavior.

NUMERICAL METHOD

The computational study was performed using a flame code already developed to study the transient behavior of C2H4/O2/N2. The transient flame responses from the ignition to extinction or steady burning were recorded. The flame code incorporated the description of detailed transport and reaction kinetics to accurately predict the burning characteristics. Because microgravity flames were strongly affected by radiation, a realistic optically thick radiation model was included to study both the kinetic and radiative extinction limits. The radiative extinction limit was a flame extinction state that does not exist without radiation. This limit was first reported by Chao et al. (1991). The flame temperature (maximum temperature within the flow) and flame location (where the maximum temperature is located) were studied versus time and the effects of given conditions including the flow conditions and radiative heat loss were analyzed. Cases with C2H4 17% O2 + 83% N2 with a mass flow rate of 1.51mg/s and 2 mg/s were ran.

ANALYTICAL RESULTS

As mention above, the analytical analysis had to be solved numerically and solutions of the T1(0) versus the Lamba2Bar ( 2λ ) for selected Lewis numbers were obtained. Figure 3 shows a baseline plot where the Lewis numbers of all the reactants are unity ( ROF LeLeLe == =1). It is worth noting that the dotted curve is an unrealistic solution and the solid curve is the true solution. The reason for this is that as Lamda2Bar increases, the reaction rate increases, making reaction more complete. This causes the flame temperature to approach the adiabatic flame temperature,

T1 (0) Vs. Lambda2bar

0

10

20

30

40

50

60

70

0 0.02 0.04 0.06 0.08 0.1 0.12 Lambda2bar

T1 (0)

Figure 3: ROF LeLeLe == =1. The dotted curve is not an

unrealistic solution and the solid curve is the true solution.

Unrealistic

Realistic

68

T1(0) vs. Lambda2bar

0 5

10 15 20 25 30 35 40 45 50

0 0.005 0.01 0.015 0.02 0.025 Lambda2bar

T1(0)

2.1=OLe

Baseline Decreasing OLe

Figure 6: Plot with varying OLe but fixed RLe and FLe l

meaning the change in temperature T1(0) approaches zero, in accordance to the solid curve and not the dotted curve. The point where the solid and dotted curves intersect is the smallest possible Lamba2Bar ( 2λ ) where the flame can still exist and not extinguish, reaching its kinetic extinction state. Beyond this point, there is no solution. Figure 4 shows curves with different Lewis numbers of fuel but all with Lewis numbers of

radical and oxidizer of unity. As shown, it can be seen that the Lewis number has a profound impact on the flame characteristics, especially the flame’s kinetic extinction point. The smaller the Lewis number, the smaller the reaction rate it can sustain. This is because as the Lewis number of fuel decreases, mass diffusion increases, increasing the reactant leakage that causes incomplete reaction. This leads the flame to extinguish at a much lower reaction rate. Although the spread from the baseline curve is not as severe as fuel’s, by varying the Lewis

number of the radical while fixing the Lewis numbers of fuel and oxidizer to unity, again, as Lewis number decreases, the flame goes extinct at a much lower reaction rate (see Figure 5). Finally, by varying the Lewis number of oxidizer while fixing the Lewis numbers of fuel and radical to unity, it is observed in Figure 6 that the kinetic extinction limit does not differ from the baseline, as seen from the previous two cases. The reason for this is because oxidizer is supplied from

the burner, which causes convection to dominate over mass diffusion.

T1(0) vs. Lambda2bar

0 5

10 15 20 25 30 35 40 45 50

0 0.005 0.01 0.015 0.02 0.025 Lambda2bar

T1(0)

Baseline

Decreasing RLe

Figure 5: Plots with varying RLe but fixed FLe and OLe

2.1=RLe

8.0=OLe

T1(0) vs. Lambda2bar

0 5

10 15 20 25 30 35 40

0 0.002 0.004 0.006 0.008 0.01 0.012 Lambda2bar

T1(0) Decreasing FLe

Baseline

Figure 4: Plots with varying FLe but fixed RLe and OLe

2.1=FLe 8.0=FLe

69

NUMERICAL RESULTS

Cases with C2H4 17% O2 + 83% N2 with a mass flow rate of 1.51mg/s and 2 mg/s

were ran and the transient flame responses from the ignition to extinction or steady burning were recorded. From Figures 7 and 8, it can be seen that extinction of the flame with flow rate of 1.51 and 2.11 mg/s occurs at 3.3 and 1.3 seconds, respectively. The extinction occurs when the radius

of the flame and temperature decreases abruptly. The difference in extinction time between the two flow rates is due to radiative heat loss. As flow rate increases, the radius of the flame also increases. After ignition, a spherical flame propagates out, due to the consumption of fuel and oxygen. As this is happening, the temperature of the flame decreases with time, due to the radiative heat loss (see Figure 8). It is known that radiation is a volumetric effect, that is, as the flame gets larger, the greater the heat loss due to radiation, and, therefore, the

lower the flame temperature. The temperature will drop to a point where the reaction can no longer be sustained, such that the flame will be extinguished.

CONCLUSION

Temperature Vs. Time

0

500

1000

1500

2000

2500

0 1 2 3 4 5 Time (sec)

Temperature (K)

1.51 mg/s

2.11 mg/s

Figure 8: Plots of transient response of flame temperature. Plots with flow rates of 1.51 and 2.11 mg/s

Extinction

Flame Radius Vs. Time

0

0.5

1

1.5

2

2.5

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Time (sec)

Flame Radius (cm)

1.51 mg/s

2.11 mg/s

Figure 7: Plots of transient response of flame radius. Plots with flow rates of 1.51 and 2.11 mg/s

Extinction

70

From the analytical analysis, it was observed that Lewis number of fuel in the ambient

has a profound impact on the kinetic extinction state. By decreasing the Lewis number of fuel, the flame extinguished at a lower reaction rate. From the numerical analysis, it was observed that flames with a higher flow rate extinguish quicker, due to the radiative heat loss experienced as the flame propagated out after ignition.

ACKNOWLEDGEMENTS

I would like to give my sincere gratitude to my mentor Dr. Beei-Huan Chao for all his

support and patience. I would also like to thank NASA and the Hawai’i Space Grant Consortium for giving me this opportunity. Special thanks to Karl Santa.

REFERENCES

Chao, B. H., Law, C. K. and T'ien, J. S. (1991). Structure and Extinction of Diffusion Flames with Flame Radiation. Twenty-Third Symposium (International) on Combustion, 523-531.

Liu, S., Chao, B. H. and Axelbaum, R. L., (2005). A Theoretical Study on Soot Inception in Spherical Burner-Stabilized Diffusion Flames, Combustion and Flame 140, 1-23.

71

INTELLIGENT SENSOR NETWORK FOR EXTREME ENVIRONMENTS

Mary Liang Department of Electrical Engineering

University of Hawaii at Manoa Honolulu Hawaii, 96822

ABSTRACT

This project is contributing to Dr Chris McKay’s research to determine if life on Mars

exists or could potentially exist. We have decided that a network of autonomous motes will aid us in detecting extreme life. Our goal this semester is to design a system to passively detect potential signs of life and finding the key molecular components, which is needed for life on Mars.

Last semester we researched Mar’s conditions, this semester we are designing a mote that would deploy chemicals on Mars. The chemicals will react with florescent lights and based on the chemical reaction, we would be able to distinguish what types of bio-signs are present on Mars. The main chip we intend to use is a micro-chip which, has built in humidity sensor, temperature sensor, light sensors, solar panels, and other qualities which would help us design the motes. The extra sensors (atmospheric and wind) will be deployed with the a bio-detector in order for us to have some in depth knowledge of the weather patterns to determine if Mars’s weather conditions are stable enough to sustain life. Proto-typing, debugging and testing the design is reserved for Fall 2006.

INTRODUCTION

While trying to contribute to Dr. Chris McKay’s work to find life on Mars, our group decided to design autonomous motes with a chip, a mesh system, and a simple bio-detector. We assigned tasks to different group members as following: Faye Yuen (electrical engineering) is in charge of choosing the hardware and the design of the power supply; Joshua Irvine (bio-engineering) is our biology expert; Tiffany Iiga (electrical engineering) is in charge of the communication between each mote and my partner for the system design; my part is the design of the overall system (with Tiffany) and assisting Joshua with the bio-detector.

SYSTEM

Our project consists of

designing passive autonomous motes to test in Mars analogue situations in hopes of one day sending a modified version to Mars. Our system will help collect meteorological data, which will allow us to understand more about Mar’s weather patters and detect life. Our overall design is to send a

Figure 1: Diagram of super mote

72

pyramid structure (refer to figure 1) that has a wireless mesh network that will make it possible to transmit data to earth. This structure is a pyramid design because we would like to have solar panels wrapped around outside of the structure. This will provide maximum sun exposure at most times and will ensure that structure will fall base down. There is a shaft in the middle of the pyramid containing gel stains. When the structure lands on Mars, the shaft will go a few millimeters into the ground and disperse the gel stains into the floor. LEDs located were the shaft is (lodged inside on the ceiling of the structure) will cause a reaction with the gel stains and will be detected by a photo diode array. The microchip, located near the LED, will be connected with the photodiode array and will collect the data to send it via satellite to earth. This whole system will be powered by the microchip’s solar panel (referred to as a internal power supply) and flexible solar panels (referred to as the external power supply) that will be wrapped around the pyramid’s sides.

PARTS

A crucial aspect of our project is choosing a chip that will meet our needs. The JN5121 is created by Jennic and is a low power, IEEE802.15.4 compliant wireless micro-controller with a fully compliant 2.4GHz IEEE802.15.4 transceiver (perfect for transmitting information wirelessly). The JN5121 has 64Kb of ROM, which allows a mesh network protocol and 96Kb of RAM, which allows support of the router and controller functions. There are 56 pins for the JN5121. There are 4 ADC and 4 DAC I/O pins called the Serial Peripheral Interface (SPI) which allows data transfer between the JN5121 and our external sensors, bio-detector, and the photodiode array.

The most passive technique to find chemical compounds in soil is by using gel stains. The gel stains are chemical which would have a binding only with a specific component, which we would be looking for. The Spyro Ruby gel stain is a stain that detects DNA. The Spyro Ruby’s chemicals would only react if there were DNA in the soil of Mars. If reaction (binding) occurs and light is shined on the chemical, the light would be absorbed and electrons in the soil and chemical compound would react and be excited to a higher energy level. Since electrons prefer to be at the lowest energy level, the electron would release a photon and drop to lower energy levels. The released photon has a different wavelength than the original LED light that was shined on the soil and chemical mixture. The different wavelength is easy to detect because it would be a different color. Knowing what the wavelength (color) of the reaction, we are able to buy a photodiode array, which would detect around the anticipated wavelength and send the information to the motes, which would be sent via satellite to a processing computer.

In the future, we would choose different gels to detect more specific and complex molecules and eventually organisms, but at this specific time, we would be focusing on the system as a whole and not only on the biosensors. This is assuming that Martian DNA is similar to the genetic makeup found on earth. We would also detect lipid membranes which is present in all cells. Lipid membranes are the cell walls that keep all of the cells components inside the cell.

Originally, we had the idea of a camera which would take pictures of the reaction and send it to the computer to test, but the camera would give us only some information at certain intervals. Using a photodiode array is a more practical approach because only the color (the reaction) is important data we need. Sending pictures takes up more memory and power to transmit. Even though the gel stains is used only once, it is hard to time the camera perfectly with the reaction. We would like to monitor the gel stains for more than the instant it requires to

73

take a photo because there may be more changes. We chose to solve this problem by using a photodiode array because it was the simplest way to detect different wavelengths. A photodetector is typically made of a silicon p-n junction. A photodetector converts optical input signals into electrical current. When there is light shining on a photodetector, electron and hole pairs are created in the semiconductor’s depletion region. When there is an electron and hole pair created, there is an induced electric field and the electron will go to the n region and the holes will go to the p region, which creates current. The current is what is detected and will determine what amount of light was shined on the semiconductor. Our group has chosen a S7585 Si PIN photodiode array made of ceramic by Semicoa company. We choose to detect the reaction by using the S7585 Si PIN, which detects wavelengths from 400 – 1100 nm (a very wide range) and also extends over the visual range of wavelengths. This photodiode array has such a large range, it is possible to use it for different gel stains to detect different chemicals and multiple times.

With this chip, we will program a sleep timer in order for the system to save energy and go into sleep mode when there is no data being taken. Since we are not continuously taking readings, the motes will not continuously communicate with each other. Therefore, we will program counters to control the information being taken and the information being sent.

With the software and hardware libraries provided by Jennic, the transceiver, timers, and the peripherals (SPI) of the JN512 are easily used and easily programmed. The language used for the microphip, communications protocol and the motes is C++. The Jennic Software Developers Kit provides debugging software, which is also one of the reasons we chose the JN5121.

FINAL DESIGN

In the block diagram of the environmental mote (refer to figure 2), we have the main component being the JN5121. The JN5121 has an internal temperature, light, and humidity sensors. Also, it has internal panels, which would charge the rechargeable battery. The power would be used only for the internal components of the JN5121. Internal components consist of the router, flash memory, microprocessor and all of the subcomponents. Added components are the external solar panels, which provides power for the atmospheric and wind sensor. The atmospheric and wind sensors will use the SPI inputs. The external battery (charged by the external solar panels) will also be used as a back up power source in case the internal power source malfunctions.

The super mote (refer to figure 3) encompasses the same system as the environmental mote, but with added features. The super mote has an extra bio-detector, which consists of a chemical (in a shaft), LED, and photodiode array. If there is life on Mars, there would a reaction when the chemical comes into contact with Mars’ soil. The reaction would be a different wavelength and would be detected by the photodiode array. Since the LED is needed to cause the reaction and the photodiode array needs more energy, there would be another power supply. The power supply would also act as a backup to the solar panel within the JN5121. The other purpose of having another power supply is to capture as much sunlight as possible and to store as much as possible to ensure longer use of the motes.

74

Figure 2: Block diagram for super mote

JN5121

Power Module

12-V NiMH Rechargeable

Batteries

12-V to 3-V Converter

Solar Panel

Microcontroller

Microprocessor

Counter/Timer

A/D Converter

Flash Memory

External Power Supply

Solar Panels

Rechargeable NiMH

Communications

Antenna

Router

Programmable Protocol

Internal Sensors

Temperature

Humidity

Light

External Sensor

Biosensor

Photodiode Array

Gel Stains

LEDs

Atmospheric Pressure

Wind

75

Figure 3: Block diagram for environmental mote

ACKNOWLEDGMENTS

I would like to thank my team group (Tiffany, Joshua, and Faye) for helping me though all the tough and confusing times of this project. I would also like to thank NASA and the Hawaii space Grant College for giving me the opportunity to do this project. I would like to thank the companies (Semicoa, Smart Dust, and Jennic) that helped us with the researching and understanding the products. Also, Dr. Kim Binsted, Dr. Tep Dobry, Dr. Edo Biagioni, Dr. Winston Su, and Dr. Daniel Jenkins and all the other professors we spoke to for helping us along the way, even though we did not have any background in the field. Last but not least, I would like to thank Professor Maria Cecilia Herrera Astua for her kindness and patience in helping me

JN5121

Microcontroller

Microprocessor

Counter/Timer

A/D Converter

Flash Memory Power Module

12-V NiMH Rechargeable

Batteries

12-V to 3-V Converter

Solar Panel

External Power Supply

Solar Panels

Rechargeable NiMH Batteries

Communications

Antenna

Router

Programmable Protocol

Internal Sensors

Temperature

Humidity

Light

External Sensor Atmospheric

Pressure Wind

76

edit this paper. Thank you for every one’s patience and guidance. With out everyone’s help, this project would not have matured to this level.

REFERENCES

Anderson, Dale. SETI Institute Principal Investigator. [Personal E-mail] 02 February 2006. Biagioni, Edo. Associate Professor, University of Hawaii Department of Information and

Computer Sciences. [Interview] 11 January 2006. Boal, Andrew. Astrobiology Postdoctoral Research Fellow, University of Hawaii Manoa.

[Interview] April 6, 2006; [Personal Email] 10 April 2006. Hamilton, Victoria. Assistant Professor, Hawaii Institute of Geophysics and Planetology.

[Interview – Teleconference] 9 February 2006. Jenkins, Daniel. Assistant Professor, University of Hawaii Department of Molecular Biosciences

and Bioengineering. [Interview] 24 January 2006. Mars Exploration Rover [On-line] 2006. http://en.wikipedia.org/wiki/Mars_Exploration_Rover,

May 2006. McKay, Chris. Planetary Scientist, Space Science Division of NASA Ames. [Personal E-mail]

06 February 2006. Smartdust [On-line] 2005. http://en.wikipedia.org/wiki/Smartdust, 08 December 2005. Su, Wei Wen Winston. Professor, University of Hawaii Department of Molecular Biosciences

and Bioengineering. [Personal E-mail] 01 April 2006. Waggoner, Alan. Director of the Molecular Biosensor and Imaging Center, Professor of

Biological Science and Biomedical Engineering, Carnegie Mellon University. [Personal E-mail]

28 March 2006. Williams, David R., Mars Fact Sheet [On-line] 2006.

http://nssdc.gsfc.nasa.gov/planetary/factsheet/marsfact.html, 01 September 2004.

Hardware Optimization of the Trigger Unit for Radio Frequency for Ultra High Energy Neutrino Detection in Antarctica

Brandon A. Merz

Department of Physics University of Hawaii at Manoa

ABSTRACT

This paper introduces the need for trigger modeling and optimization in the ANITA project. The assembly process, hardware components, and software architecture of the system designed to study the trigger is the

focus of the paper. An analysis and discussion of the project is presented.

INTRODUCTION – TRIGGER MODELING This project dealt with the optimization of trigger rates, where a trigger is defined

as any number of combinations of hits; a hit being an electro-magnetic pulse captured by the antennas that exceeds a predetermined signal threshold, thus notifying the system of an event of interest.

In order to improve the GZK (Greisen-Zatsepin-Kuzmin) neutrino reach of ANITA (Antarctic Impulsive Transient Antenna) it is essential to operate as far down into the thermal noise as is feasible [1]. From a practical perspective, this limit is set by the rate for accidental triggers that the data acquisition system can tolerate. A useful picture for understanding the level 1 trigger is provided below in Fig. 1. Here one of the many Quad-Ridged horns on the fully assembled ANITA payload, Fig. 2, is shown. The requirement of 3 of 8 hits is a way of reducing accidental thermal-noise triggers to a rate of less than 5 Hz for the final L3 trigger rate, which is based upon spatial and temporal L1 coincidences.

Fig. 1: A diagram of a single antenna trigger path, where vertical and horizontal polarizations are combined into right and left circular polarizations and subbanded, from which we form a level 1 trigger.

Fig. 2: Physical partitioning and definition of the L2 and L3 trigger segments based on the 16-fold symmetry of the ANITA antenna array, as seen on left. On the right is a sample of the trigger geometry considered for a given Phi (1 of 16) sector. The Level 1 trigger rate is represented by Eq. 1 below, where a Level 1 trigger

requires 3 of 8 hits for individual sub-trigger rates rN within a certain timing interval τ:

Eq. 1: ( )2383 ! 5 ! 3

! 8 τNof rR ⋅⋅

=

This rate and expectations for subsequent coincidences for the noise have been

assumed to be of a Gaussian noise form for the ANITA project, but a more realistic 1/f + white noise form will be studied in order to model observed noise phenomena after verification of a working white noise signal.

FABRICATION AND ASSEMBLY The reasons for choosing the SPAR (SalSA Transient Prop Askaryan Receiver)

board, shown on the next page in Fig. 3, for this project will be discussed in a later section, meanwhile this section will focus on the assembly of the board [2].

Fig. 3: The assembled SPAR board showing 3 signal inputs on the left, a USB2 port on the bottom right, and two DACs directly above the XILINX FPGA. The SPAR board was designed in the IDLab previously and uses surface mount

components. The PCB (Printed Circuit Board) itself was manufactured by Sierra Proto Express based on a design by Dr. Varner and Larry Ruckman. Assembly was done by the author in the lab using soldering stations equipped with microscopes necessary to assure accuracy in part placement.

PROGRAMMABLE NOISE SIMULATION CIRCUIT

Below is a sequence flow describing the interaction between the data generation,

the digitization, and the evaluation segments. This figure will be discussed throughout the paper.

Fig. 4: SPAR Data Flow showing the interaction between the PC, SPAR board, and output.

Creating a signal that accurately represents the trigger impulse being received is difficult using analog methods. The signals can be reproduced digitally in a simpler and more versatile fashion. The SPAR board was a suitable choice for this project because it had the necessary IO interfaces to communicate with a PC, store information, and output an analog signal. The utilization of the board will be described in subsequent sections.

This project began in the PC with a C++ algorithm, which generates white noise data to be sent to the SPAR board where it interacted with the USB controller [3]. A histogram of the noise generated is shown below in Fig. 5. The x-axis range of 8 is subdivided into 1000 bins where the generated numbers are entered stored based on their value. The y-axis shows how many of the numbers are in the corresponding bins. A time series of this noise is shown in Fig. 6.

Fig. 5: Histogram of generated Gaussian noise with

a mean of 0 and variance of 1.

Fig. 6: Time series of the Gaussian noise.

-3

-2

-1

0

1

2

3

Time

Volta

ge

The sequence of events leading up to the PC output is illustrated in Fig. 7. The

first segment is a C++ program to produce white noise data i.e. ‘Noise Generator’. This output can then be either transferred from the PC to SPAR module or read into a FFT (Fast Fourier Transform) module to plot the transform in order to check for a flat distribution, indicating equal power distribution over all frequencies for the initial white noise data. A numerical transform is available for the discrete data used in this project and was used as the transform module [4].

Modular design allows any text file of data to be loaded from the PC to SPAR, so that once a working system has been verified the only task remaining would be to produce a desirable list of data.

Fig. 7: PC Data Flow

In order for this data stream from the computer to the SPAR board to be stored in

FPGA (Field Programmable Gate Array) memory the EEPROM (electrically erasable programmable read only memory) was programmed using a JTAG header. This program loaded instructions into the FPGA upon power up, which allowed data to be stored before being sent to the DAC. The particular programmable logic device being used for this project was the Xilinx Spartan 3 FPGA [5].

At this point the DAC (digital to analog converter) receives information in bursts of 2 KB of data. This signal can then be output to a simple RC integrator circuit consisting of a 75Ω resistor and a .1nF capacitor to remove the frequencies above 20 MHz. Loss of the high frequencies is acceptable as only the lower frequencies are of interest. After the signal has passed through the low-pass filter it could be read and recorded by an oscilloscope or input back into the SPAR board via the LAB3 (Large Analog Bandwidth Recorder and Digitizer with Ordered Readout) chip [6]. The oscilloscope would sample the signal based on a trigger, according to the Nyquist theorem in order to accurately represent the waveform.

In order to verify a working board and simulation the white noise output will be tested against a white noise function on a function generator. Once a signal that is equivalent to the noise diode signal has been produced the implementation of a 1/f + white noise signal will be completed. The white noise and 1/f + white noise signals are

shown below in Fig. 8 A and B respectively. The lower limit for the bandwidth is determined by the DAC speed and the upper limit by the low pass filter.

Fig. 8: White noise power spectrum shown in A) white noise +1/f shown in B)

CONCLUSION

The project was not able to be completed according to schedule as there were problems in the programming interaction between the USB and SPAR board. The project will be continued to completion and once basic operation has been verified the 1/f addition will be modeled.

ACKNOWLEDGEMENTS

The author would like to thank the Hawaii Space Grant College for the funding to

pursue this project. Thanks to Dr. Varner for mentoring the project and to the IDlab staff for help with many of the technical concerns.

REFERENCES

[1] The ANITA Trigger Logic: Estimates of Thermal Noise Trigger Rates and Practical Operating Thresholds, G. Varner et al., available online: http://www.phys.hawaii.edu/~idlab/publications/ANITA_GTM_Note.pdf

[2] http://www.physics.ucla.edu/astroparticle/salsa/slacfeb05/varner_SalSA_digitize.pdf [3] http://www.everythingusb.com/usb2/faq.htm [4] Numerical Recipes in C++, Press & Teukolsky; ch. 12 [5] http://direct.xilinx.com/bvdocs/publications/ds099.pdf [6] http://www.phys.hawaii.edu/~idlab/publications/LABRADOR.pdf

83

EFFECTS OF LOW REYNOLDS NUMBERS ON THE AERODYNAMICS OF MICRO-AIR VEHICLES

Shelly A. Migita

Department of Mechanical Engineering University of Hawaii at Manoa

Honolulu, HI 96822

ABSTRACT

Unmanned aerial vehicles (UAVs) come in various sizes and have several benefits. They are more compact, portable, capable of working in hazardous conditions, and cost effective than larger aircrafts, which often require the control of pilots inside of cockpits. Due to their novelty, there is still much research that must be done in this area. Topics of research include the interplay between low Reynolds numbers and aerodynamic loads, such as lift and drag, and flow characteristics that contribute to such loads. A computational fluid dynamics (CFD) program, COMSOL Multiphysics, was used in the research. This research effort showed the characteristics of a steady, low-Reynolds flow over a stationary, elliptical airfoil. As angle of attack increased, vortices developed at the trailing edge of the airfoil. For Reynolds number of 150, the vortices remained attached to the airfoil. However, for high angles of attack and Reynolds numbers around 1000, the vortices that formed at the trailing edge were shed into the wake region. Higher angles of attack also appear to correlate with vortex generation and shedding from the leading edge as well as the trailing edge of the airfoil. The lift to drag ratio generated by such stationary airfoils, however, did not appear to be sufficient enough to sustain MAV flight. The research shows that a simple, sinusoidal flapping mechanism and elliptical airfoil has the potential to generate a sufficient lift to drag ratio, but requires further investigation.

INTRODUCTION

Unmanned aerial vehicles (UAVs) are like jets minus the pilots in the cockpits. UAVs come in various sizes and have proven particularly beneficial in areas such as intelligence, surveillance, and reconnaissance (ISR) and communication. Some UAVs are so small that they can be disassembled into pieces small enough to fit into a backpack. In addition to being compact and portable, UAVs, particularly on a smaller scale, are also more cost effective and have the ability to work in hazardous conditions. Due to their novelty, there is still much research that must be done in this area. The low-Reynolds effects on the aerodynamics of small UAVs, in particular, require greater attention. Topics in this area include the interplay between low Reynolds numbers of a smaller classes of UAVs, called micro-air vehicles (MAVs), and aerodynamic loads, such as lift and drag, and flow characteristics that contribute to such aerodynamic loads. This research investigates the effects of low Reynolds numbers on the aerodynamics of MAVs using a computational fluid dynamics (CFD) program, COMSOL Multiphysics. The findings of this research effort will not only enhance our understanding of the aerodynamics of these small airplanes, but they will also contribute to the design of novel, improved, and ultimately more reliable MAVs.

84

METHODS

COMSOL Multiphysics was used to model the fluid flows over stationary, elliptical airfoils. Since the research pertained to low Reynolds numbers such as those found in the flight of insects, the elliptical airfoil was selected as a means of providing a simplified, but similar geometry, to the cross section of an insect wing. The flows over stationary, elliptical airfoils were modeled for angles of attack between zero and twenty degrees, with respect to horizontal, and Reynolds number of 150. The lift and drag coefficients, CL and CD, and lift to drag ratios were calculated for each of twenty models. The optimal angle of attack of the cases where Re = 150 was the case that produced the highest lift to drag ratio. In an attempt to increase the resulting lift to drag ratio to a more desirable value, select angles of attack beyond twenty degrees were modeled for an unsteady flow and Re = 1000. Due to the unsteady flow, the models were analyzed with respect to time. The lift and drag coefficients and subsequent lift to drag ratios were then calculated for each time step and analyzed with respect to non-dimensionalized time. An elementary flapping mechanism was also investigated. Due to the complexity of the transient model, the accuracy the researched model was based upon research performed by Dr. Jane Wang in “Vortex shedding and frequency selection in flapping flight.” In the research, a 40 x 40 hyperbolic-elliptic grid and finite element method was used in the analysis.

RESULTS

The lift to drag ratios obtained from each of the twenty, stationary airfoils showed that the stationary wing with a seventeen-degree angle of attack resulted in the highest lift to drag ratio. This lift to drag ratio was 1.4, which, by the Defense Advanced Research Projects Agency (DARPA) standards, is not high enough to sustain MAV flight. Figure 1 shows the pressure

distribution and streamlines that were modeled for the stationary, elliptical airfoil with a seventeen-degree angle of attack. The dark region under the leading edge at the left tip of the airfoil shows a region of high pressure developing just under the airfoil. The lighter area in the wake region behind the airfoil shows a region of lower pressure. The streamlines plotted in the figure shows the beginnings of separation in the wake region.

85

Unlike the flow characteristics observed in Figure 1, the flow characteristics exhibited by the unsteady, stationary, elliptical airfoil was quite different. Figure 2 shows a screen shot of a stationary, elliptical airfoil with a thirty-degree angle of attack. Re = 1000, and unsteady air was modeled over the airfoil. The model shows dark pockets of low pressure in the wake region that get progressively lighter as the distance from the trailing edge of the airfoil increases. These regions of low pressure were vortices that formed, detached from the airfoil, and flowed through the wake region. The detached vortices are shown in the figure as lighter-colored orbs. Unlike the previous figure, Figure 2, vortices also began to develop at the leading edge of the airfoil. The formation and detachment of vortices resulted in fluctuations in the lift and drag coefficients and lift to drag ratios. The resulting average lift to drag ratio for the case shown in Figure 2 was about 1.2, and the maximum lift to drag ratio was 1.7. In order to increase the lift to drag ratio, a simple flapping mechanism was modeled for the elliptical airfoil in an unsteady flow and with Re = 1000. Figure 3 shows a screen shot of a simple sinusoidal flapping mechanism and flow characteristics exhibited by the model. Figures

4 and 5 show the resulting changes in the lift and drag coefficients with respect to time. The results in Figures 3 through 5 were obtained using Stc = 1.0, Sta = 0.16, u0 = 1, and c = 2. The flapping motion of the elliptical airfoil is governed by the equation,

u1 t( )= 2πfAsin 2πft( ), where f is the flapping frequency of the airfoil.

86

DISCUSSION

While stationary airfoils do not appear to generate a sufficient lift to drag ratio, the flapping airfoil, such as the one shown in Figure 3, provides a baseline from which other studies may be performed to determine the flapping mechanism necessary to produce a sufficient lift to drag ratio. Fruit flies, for example, employ highly complex flapping mechanisms in order to maneuver properly. These flapping mechanisms often involve wing rotation, an addition flapping mechanism beyond the scope of this research. Thus, significant research must still be completed in the rewarding and promising area of MAV aerodynamics.

CONCLUSION

Based upon the results, it appears as though there is a range of angles of attack during which the vortices that form remain attached to the body. After a certain critical angle of attack, the vortices are shed off of the airfoil and through the wake region. As shown in Figure 2 and 3, vortices also begin to form at the leading edge of the airfoil with increasing angle of attack. The flapping frequency may be a factor in the vortex formation about the airfoil. A higher flapping frequency would likely cause vortices to be shed at a higher rate. In order to sustain flight, however, it appears as though a stationary wing will never be sufficient. The simple flapping mechanism modeled in the research will likely require modification to avoid generating a negative lift caused by an upward stroke, identical, but opposite, to that of the downward stroke. It is possible that by modifying the flapping frequency and rotation of the airfoil might help to generate a sufficient lift to drag ratio. Ultimately, much more research must be done in order to understand the complexities of low-Reynolds aerodynamics.

ACKNOWLEDGEMENTS

The author would like to thank NASA and the Hawaii Space Grant Consortium for allowing undergraduates to pursue research in their chosen fields. Special thanks also go out to mentor, Dr. Marcelo Kobayashi, and Hugo Pedro for all of their time, patience, guidance, and assistance throughout the course of the research. Dr. Marcelo Kobayashi, in particular, provided immense guidance regarding the direction of and analysis in the research. Finally, mahalo to

87

Brent Uyehara for providing computing power, stress relief, and incredible patience and support. This research would not have been possible without the support of such kind and dedicated individuals.

REFERENCES

Defense Advanced Research Projects Agency (DARPA) Defense Sciences Office (DSO). Proposer Information Pamphlet for Defense Advanced Research Projects Agency (DARPA) Defense Sciences Office (DSO) Nano Air Vehicle (NAV) Program. BAA 06-06, 7. Wang, Z. Jane. (1999) Vortex shedding and frequency selection in flapping flight. J. Fluid Mech. 410, 323-341.

88

LAVA FLOWS IN THE THARSIS REGION OF MARS: ESTIMATES OF FLOW SPEEDS AND VOLUME FLUXES

Carolyn Parcheta

Department of Geology and Geophysics University of Hawai’i at Manoa

Honolulu, HI 96822

ABSTRACT

The goal of this study was to characterize the eruptive behavior of Olympus Mons, Mars, by estimating flow speeds and volume fluxes for lava flows produced from various vents on the volcano. During my fellowship, I have made width, depth and ground slope measurements of 65 lava tubes and 271 channels from 56 MOC images and 65 THEMIS images of Olympus Mons. The data show that channels are more numerous than tubes, but this may be due to the fact that tubes cannot be detected until they collapse and are therefore likely under represented due to sampling biases. We observe that tubes are prevalent near the summit while channels are dominantly seen on the middle and lower flanks as well as on the basal scarp. Channels are found on a variety of slopes from 0 to 29°, while tubes are found on slopes less than 10°. This is consistent with the terrestrial experience that tube roofs form less readily on steep slopes.

INTRODUCTION

In order to determine and understand the eruptive behavior of Olympus Mons, theory and photogeology must be utilized. Theory refers to applying present knowledge of Earth’s volcanic eruptions, especially the current eruption of Kilauea, to images received by Martian satellites so that hypotheses about volcanism on Mars can be made. Many features that are seen on Olympus Mons are also seen on volcanoes on Earth, and thus photogeology refers to interpreting the geology of Mars by comparing Martian images with terrestrial features. With the application of these two approaches, the collective scientific knowledge of

geologic processes that occurred, or are occurring, on other planets will be enriched and the evolution and history of bodies in our solar system will become clearer. Specifically, this research was aimed at estimating flow velocities and volume fluxes for lava on Olympus Mons (based on both known and assumed input parameters) and to characterize lava emplacement styles. Lava morphologies that I looked for are lava tubes, shown in Figure 1 and Figure 3, and lava channels, shown in Figure 2 and Figure 3. In the process of doing this, I examined the data for correlations and trends among variables such as velocity, flux, emplacement, location on the volcano, and topographic slope.

Figure 1: Lava tube, photo by Scott Rowland

89

METHODOLOGY

The methodology used for this research

involves four main steps. First, I identified pristine lava tubes and channels from the data received by the Thermal Emission Imaging System (THEMIS) on Mars Odyssey and the high-resolution Mars Orbiter

Camera (MOC) aboard Mars Global Surveyor as demonstrated in Figure 3. It is important that these lava tubes and channels are pristine and crisp so that shadow measurements and width measurements may be as accurate and precise as possible. The resolution of the MOC images that were used was three meters per pixel while the resolution of the THEMIS images that were used were either 18 or 36 meters per pixel.

Second, I processed and reprojected these images using the UNIX program ISIS in order to successfully compare them with Mars Orbital Laser Altimeter (MOLA) digital elevation data. The ISIS command thm2isis, thmvismc, levgeoplane and cub2envi were used for THEMIS images and moclevall was used for MOC images. Detailed information on these commands is available from http://isis.astrogeology.usgs.gov/documents/ Isis2Tutorials/index.html. As a final processing step I exported the images in a format compatible with ENVI, the software used to make the measurements.

Third, I measured widths and depths of channels and tubes from the reprojected images and I also measured the ground slopes from a slope map that was produced based on the MOLA 1/128° DEM. Widths were measured

by averaging the distance of two lines placed orthogonal to the channel walls. One line was between the top edges, representing the largest possible width of the tube or channel and one line was placed on the gray area at the bottom of the tube or channel, representing the smallest possible width. Depth measurements were made by measuring the length of dark areas inside the lava tubes and channels and dividing by the tangent of the incidence angle in radians. These dark areas were assumed to be shadows, but were compared to the darkest areas of the image they originated from to make sure that what was measured actually was a shadow. Ground slopes were taken by comparing the MOC and THEMIS images to the MOLA slope map, and the slope of corresponding area on the MOLA slope map between the two images gave the slope of the ground surface underlying the lava.

Lava flow

Lava tube

Lava Channel

Figure 2: Lava channel, photo from Scott Rowland

Figure 3: Martian lava flow morphologies in THEMIS image V11326014

90

Figure 4: Percent LTC of 336 Measured

81% 19%

Channels Tubes

Finally, lava flow velocities and total volume fluxes, based on assumed values of viscosity and density, were calculated. The Jeffreys equation for laminar flow (Jeffreys, 1925),

which is

u = g sinθρh2

3η, was used for velocity and the subsequent resultant velocity was

multiplied by the cross-sectional area of the tube or channel to get the volume flux. When calculating the velocity, g is the gravitational acceleration of Mars – which is 3.73 m/s2, θ is the ground slope that was measured from the MOLA slope map, ρ is the assumed lava density of 2000 kg/m3, h is the flow thickness (which is calculated as 75% of depth for channels and as a function of (-0.08)*ground slope +1 for tubes) and η is the assumed flow viscosity of 3000 Pas. It is important to note that estimates for lava flow velocities and lava volume fluxes incorporate a couple of assumed input parameters. The flow density and dynamic viscosity are assumed for this research project because there are no samples to inspect or methods to determine them from Earth. The values chosen reflect mean numbers for terrestrial basalts.

RESULTS

The data collected reflect 336 lava

tubes and channels from 121 MOC and THEMIS images. Of these lava tubes and channels, 81% were lava channels and 19% were lava tubes, as seen in Figure 4. However this does not necessarily imply that lava channels are more abundant than tubes on the Marian surface. Instead, it may reflect a sampling bias since lava tubes cannot be detected until the roof collapses in enough areas to trace the tube. There is a correlation between the emplacement of tubes and channels with the ground slope on which they occur. A ground slope of 10° appears to be the cutoff for tube formation. At or above a ground slope of 10°, only channels are observed.

TTHHEEMMIISS DDaattaa::

WWiiddtthh == 2244..33 –– 332233..99 mm DDeepptthh == 11..44 –– 4422..88 mm ((MMeeaann == 111188..00 mm)) ((MMeeaann == 99..55 mm))

uu == ~~00 –– 3344..99 mm//ss QQ == 44..33 –– 8833,,660000 mm33//ss ((MMeeaann == 44..11mm//ss)) ((MMeeaann == 66,,553322..55 mm33//ss))

MMOOCC DDaattaa::

WWiiddtthh == 77..44 –– 339900 mm DDeepptthh == ..22 –– 4477..11 mm ((MMeeaann == 7700..11 mm)) ((MMeeaann == 88..22 mm))

uu == ~~00 –– 5566..22 mm//ss QQ == ~~00 –– 112288,,220000 mm33//ss ((MMeeaann == 44..66 mm//ss)) ((MMeeaann == 44,,991166..11 mm33//ss))

Totals:

Width = 7.4 – 389.5 m Depth = 0.2 – 47.1 m u = ~0 – 56.2 m/s Q = ~0 – 128,200 m3/s (Mean = 94.1 m) (Mean = 8.8 m) (Mean = 4.4 m/s) (Mean = 5,724.3 m3/s)

Figure 5: Numerical results from MOC and THEMIS.

91

Diverse data showed a wide range of results in terms of width, depth and slope measurements along with flow velocity and volume flux calculations. However, the ranges for MOC derived measurements and the ranges for THEMIS measurements correlated almost exactly. In short, both data sets support the same ranges of lava tube and channel widths, depths, slopes, velocities and fluxes as shown in Figure 5. The means however, differ for each data set. This is most likely due to the image resolution and how much area is covered in the image. MOC images gave means that were lower than THEMIS (Figure 5).

The calculated velocities and flux rates show overlap with terrestrial values for basaltic volcanoes. However, the upper ends of the Martian ranges are much higher than what is recorded on Earth. This could be for two reasons: Martian gravity and the size of Olympus Mons. Gravity on Mars is roughly one third of gravity on Earth, and this would allow magma to reach the surface of Olympus Mons easily and readily. Also, dikes that feed the flank eruptions could occur more numerously and have greater widths (Wilson and Head, 1994) than terrestrial volcanoes, thus providing more magma to the surface and enlarging the effusion rate. Since Olympus Mons is the largest volcano in the solar system, it would most likely have an extensive magma chamber that would have the capacity to erupt large amounts of lava (Zuber and Mouginis-Mark, 1992)

RESULTS: ERROR ANALYSIS

I believe my width and slope measurements are fairly accurate. Probably, the largest

source of error is in the depth measurements because of the slight ambiguity of where a shadow ends in a picture. There is possibility for both over- and under- estimation depending on the contrast setting during each measurement. In addition, overestimation can occur when surges in flux overflow the channel of transportation and build up levies higher than the original channel. Alternatively, fast velocities on steeper slopes could promote lava downcutting into the volcano, again making the channel deeper than it was upon formation. Also, it is possible that some dark areas, such as in fissures or cracks, were misidentified as shadows of lava tubes and channels and thus lead to erroneous depths. Finally, an error may lie in using the wrong velocity equation for a few lava tubes and channels with high Reynolds numbers. For Reynolds numbers over 2300, a velocity equation for turbulent flow should be used, not laminar flow.

CONCLUSION

Martian volcanism appears to resemble basaltic volcanism on Earth. Lava with low viscosities appears to have built up Olympus Mons, which is the largest volcano in the solar system. Slopes are relatively shallow due to this low viscosity and thus Olympus Mons is a shield volcano. The mean widths, depths, effusion rates, and flow velocities seem consistent with those observed on Earth, even though the upper ends of the ranges are much higher than anything recorded in terrestrially. There is a negative correlation between emplacement style of lava tubes and a positive correlation of channels with both ground slope and velocity: as slope and velocity increase, tube occurrences decrease and channels increase in number. There is a sampling bias because tubes cannot be detected until there is a structural failure of the roof, however we were able to document enough of them to make good estimates of flow velocities and fluxes through them. I adopted strict criteria for using pristine and crisp lava tubes and channels, however, the data set could be expanded later to include features not as well preserved.

92

FUTURE WORK

This project could be enhanced by adding data from other Tharsis Volcanoes (Arsia Mons, Pavonis Mons, and Ascraeus Mons) on Mars, but the data for Olympus Mons is well covered from the MOC and THEMIS images of this project. With enough data from all four volcanoes, it would be possible to compare and contrast effusion rates and volume fluxes to get a broad picture of volcanic activity from the Tharsis area of Mars.

ACKNOWLEDGEMENTS

I would like to thank the NASA and the Hawai’i Space Grant Consortium for both supporting and funding this wonderful and amazing research opportunity. I would also like to thank my parents, Debra and Mark Parcheta, and my younger brother, Scott Parcheta, for supporting me and encouraging me along the way through both this project, my school work and college life in general. Finally, my biggest thanks go to my mentors, Sarah Fagents and Barbara Bruno, for letting me do research, learn about new topics and grow with this project. Your advice and guidance is always greatly appreciated and your enthusiasm for doing science as well as your amazing trips for field work is truly my inspiration. Mahalo nui loa.

REFERENCES

[1] Jeffrey’s, H.J., (1925)The flow of water in an inclined channel of rectangular section, Phil. Mag., 49 (6), 793-807. [2] Wilson, L., and J.W. Head, (1994) Mars: Review and analysis of volcanic eruption theory and relationships to observed landforms, Rev. Geophys., 32, 221-264. [3] Zuber, M.T., and P.J. Mouginis-Mark, (1992) Caldera subsidence and magma chamber depth of the Olympus Mons volcano, Mars, J. Geophys. Res., 97, 18,295-18,307,.

93

MAPPING THE PREHISTORIC STATUE ROADS ON RAPA NUI USING REMOTE SENSING SATELLITE IMAGERY

Gabe Wofford

Global Environmental Science University of Hawai’i at Manoa

ABSTRACT

In an extension of the work of Drs. Hunt and Lipo published in Antiquity (2005), I

surveyed the prehistoric roads of Rapa Nui (Easter Island) using ArcGIS software and satellite images of the island. A pre-determined set of criteria was utilized, though there was much potential for error with both false positives and negatives. A great deal of progress was made in identifying potential locations of prehistoric roads. This survey only provided possible sites, however, which will require confirmation through ground survey in future visits to the island.

INTRODUCTION

This project served as a continuation of the work of Dr. Carl Lipo of California State

University Long Beach and Dr. Terry Hunt of the University of Hawai’i-Manoa in mapping the prehistoric statue roads of Rapa Nui (Easter Island). As published in Antiquity, their article presented “an extensive analysis of the island’s roads as a complement to the intensive studies undertaken by” Dr. Charlie Love (Lipo and Hunt 2005). Dr. Love has recently researched the composite features of the ancient roads, including cross-sectional excavations (Love 2000). The documentation of the roads is an essential element in the current debate surrounding models for movement of the monolithic stone statues of Rapa Nui (Lipo and Hunt 2005). Using satellite imagery analysis and an established set of criteria, the island was surveyed for potential road features which will require further ground-truthing for confirmation. I was able to expand the catalog of prospective ancient roads beyond that established by Lipo and Hunt.

METHODS

The satellite images consist of three panchromatic images acquired by the QuickBird

satellite in December 2001 and February 2002. The images account for 92% of the island’s surface at a resolution pixel size of 70 cm. Taking into account the inability to read the ground topography in the modern town of Hanga Roa or beneath tree canopies and cloud cover, approximately 85% of the island is left available for examination. Extensive agricultural practices in the central part of the island also made survey of roads difficult in those areas. The images were useful due to the fact that their resolution is “an order of magnitude greater that the width of the road features” (Lipo and Hunt 2005).

The establishment of criteria for this search started with consultation with Drs. Hunt and concerning the original survey methods for the preliminary roads search. Potential road features can be recognized as linear forms on several criteria, often found in combinations, including vegetation differences, depressions filled with cobble scree, banks, trails between statues, erosion patterns and shadow marks. These features tend to display themselves as chromatic variations in the satellite images (Figure 1). Following patterns of statues across the island with an overlay

94

layer showing moai (statue) locations proved very useful in furthering the roads survey. The moai can be seen to create a somewhat linear configuration across the landscape. The location of these statues away from ahu (ceremonial platforms), indicates they were left en route, and never completed their journeys, thus lying along the ancient roadways.

Figure 1. A panchromatic 70cm resolutionQuickBird satellite image showing an ancient road section leading west-south-west from the Rano Raraku statue quarry (A). Statues that surround the quarry are easily visible in this image (B) with modern tourist trails among them, as is the modern parking lot and modern road (C). The ancient road (D) is visible primarily as a horse trail and as a line of vegetation that runs from the north-east to the south-west corner of the image. This feature likely reflects sediment compaction with greater water retention and subsequent vegetation growth. Multiple large statues (moai) line this road near the quarry (E). The satellite image was provided by RADARSAT, Inc and DigitalGlobe, Inc (Lipo and Hunt, 2005).

Primarily I had to accustom my eye to recognizing such features on the images. Beginning with a blank slate, I retraced the roads shown by Hunt and Lipo (2005) (Figure 2). The original paper confirmed 32 km of roads on the island (Figure 2). After following and extending these features, I sought to track linear features branching out from the primary moai

95

quarry at Rano Raraku crater. A systematic survey by transects across the island revealed more sites for confirmation that could not be seen in relation to the central quarry.

Figure 2. An island-wide distribution of 702 statues (moai, yellow circles), 87 topknots (pukao, blue triangles) and confirmed ancient statue roads (red lines) made on a mosaic of satellite images. About 32 km of roads are shown. The north-north-west road extends 2.7 km; the west-north-west road, 4.5 km, with a western branch an additional 2.6 km; the west-south-west road (discontinuous) 4.0 km; the main southernmost road 8.6 km; Rano Kao Crater road 3.6 km; and the west-north coast road, 13.0 km. A possible road leads directly north from the quarry at Rano Raraku (4.4 km), but additional field evaluation is necessary to confirm an ancient road given historic and modern activities on the same route (Lipo and Hunt, 2005).

Difficulties in the survey included both false positives of other linear features besides roads, and false negatives where roads were obscured or confused for modern objects. Historic and modern roads as well as stone walls appear clearly as linear objects on the image, but can usually be eliminated as possible roads. However livestock trails and dry streambeds and erosion channels (there are no permanent streams on Rapa Nui) present similar chromatic signatures to known road features. Overlap occurs in many areas where modern roads and hiking or livestock trails follow prehistoric pathways. When development first began on the island, the easiest routes for travel were along the previously established prehistoric roads. With the inconsistent, rocky landscape of the island, the easiest paths for jeeps and horses were those

96

cleared for statue movement. It is difficult to distinguish places where this occurs. Roads are also undetectable in developed areas. This difficulty primarily occurs in the southwestern region of the island, where the population is concentrated in the lone city of Hanga Roa and surrounding farmland.

Following the initial survey, I compared the satellite image to historic surveys and existing field school data. A map created by French explorers in 1877 shows several paths and roads on a distorted image of the island. Despite a measure of distortion, this map is useful in that it illustrates road patterns before ranches obscured large portions of the island and livestock ran trails across the landscape. A second map by the Heyerdahl expedition of 1955-56 shows several known prehistoric roads as well as jeep roads on the island. This survey represents the island before most of the modern roads were built, and offers some clue as to the historic and modern use of prehistoric roadways. However at this time the issue of overlap presented itself with the need to acknowledge that historic and modern trails likely followed ancient roads. Lastly, I used coordinates collected by past University of Hawai’i field schools on Rapa Nui to distinguish features which have been investigated and deemed to represent prehistoric statue roads.

RESULTS AND DISCUSSION

I was able to extend potential sites for future confirmation beyond the survey of Hunt and

Lipo. This is evidenced in the extension of the road along the south coast running parallel to a modern paved road. The eastern portion of the island to the northeast of Rano Raraku crater and around the Poike peninsula also contained several possible roads (Figure 3). Unfortunately, this area is one of the least open to survey due to private ownership and uncertain terrain.

Documentation of the ancient roads of Rapa Nui allows for the evaluation of competing models for how the statues were moved, one of the great mysteries of Rapa Nui. A record of the roads provides the minimal distances that the statues could be transported (Van Tilburg 1994). Mapping the paths can also provide clues for the social, economic and political organization of the ancient islanders. The roads emerge from the quarry in a radial pattern, suggesting that they were not necessarily shared, but each region (potentially related to individual social groups) had its own road for delivery of statues from Rano Raraku. This apparent independence and lack of cooperation indicates the lack of a centralized authority. A model of smaller, competing groups is more likely.

97

Figure 3. The south coast road and eastern portion of the island exhibit the most development following my survey. These sites would be the most beneficial to survey for confirmation by later field schools.

ACKNOWLEDGEMENTS

I would like to thank Drs. Hunt and Lipo for their help and cooperation in aiding my

survey and intial contributions. Also, Alex Morrison was very helpful in my familiarization with ArcGIS software used for analysis.

REFERENCES

Lipo, C.P. and Hunt, T.L. (2005). Mapping prehistoric statue roads on Easter Island. Antiquity 79, 158-168.

Love, C. (2000). More on moving Easter Island statues, with comments on the NOVA program. Rapa Nui Journal 14, 115-118.

Van Tilburg, J.A. 1994. Easter Island archaeology, ecology and culture. London: British

Museum Press.

98

INTELLIGENT SENSOR NETWORKS FOR EXTREME ENVIRONMENTS

Faye S.Y. Yuen Department of Electrical Engineering

University of Hawai’i at Manoa Honolulu, HI 96822

ABSTRACT

A network of autonomous motes that are capable of detecting life in extreme

environments may hold the key to finding habitable environments beyond Earth. Laying the foundation is Dr. Chris McKay’s ongoing research to detect life in Mars and Europa analog environments. The primary goal of this investigation has been to find signs indicative of life and to design a device that will collect such data. Each mote will be designed to resist extreme environmental conditions (ex. hot and cold temperatures), provide a long term means of data collection by keeping power consumption to a minimum, and provide relevant and accurate data. Characteristics that differentiate this project from life-detecting rovers (such as the Mars Pathfinder or Viking) are the motes’ size, ease of deployment and operation, low cost of production, and low environmental impact.

INTRODUCTION

By creating a network of autonomous motes to be deployed in extreme environments (i.e. those similar to conditions on Mars), this project aims to assist Chris McKay, PhD, in his research, and also meet Goal 5 of the NASA Goals and Objectives from the NASA Strategic Plan, Appendix III, which is to "Explore the solar system and the universe beyond, understand the origin and evolution of life, and search for evidence of life elsewhere."

By researching life in extreme environments on Earth, we can come closer to determining whether life, as defined by Earth standards, exists, had existed, or can exist on Mars. This issue will be fundamental to understanding our own origins and evolution, and answering the universal question, "Are we alone in this universe?"

An initial network of three motes equipped with sensors capable of detecting some of the important conditions for life will serve as the means of collecting data. A mote is defined in Wikipedia as a “device containing sensors, computing circuits, bi-directional wireless communications technology and a power supply”.

The three motes will consist of one “super-mote” and two “environmental motes.” The super-mote will collect specific indicators of life (i.e. lipid membranes and macromolecules including proteins, DNA, and carbohydrates). Environmental motes will measure specific characteristics such as temperature and humidity that complement the data collected by the super-mote.

METHODS

This project spans 3 semesters: Fall 2005 (Research), Spring 2006 (Design), and Fall 2006 (Build and Test). The mote network and its current design is based on prior research and feedback from experts in various fields.

99

The technical specification of each mote is designed to meet the anticipated environmental conditions that are expected to exist on Mars. To determine such conditions, we examined those Mars analog environments found on Earth (i.e. Chile’s Atacama Desert, and the Arctic Dry Valleys), the previous conditions that the earlier Mars rovers encountered, and consulted with Dr. Victoria Hamilton, University of Hawaii’s resident “Mars Expert.”

Critical in this study is the choice of measurable indicators to determine what is life and how to detect it. Through meetings and emails with Dr. Andrew Boal and Dr. Chris McKay, it was decided that the mote network should detect lipid membranes (key indicators of life), and corresponding temperature and humidity to provide complementary information on the environmental conditions that allow that organism to survive. Dr.’s Dale Anderson, Daniel Jenkins, Wei Wen Su, and Alan Waggoner provided the needed insight into detecting such biological life signs. This bioenvironmental analysis led to the choosing of chemical gels as the most viable means of bio-detection.

The Zoë Mars rover, developed by the researchers at Carnegie Mellon University, has previously used chemical gels to detect organic molecules in the Atacama Desert. Spraying the dyes on the ground cause organic molecules to fluoresce. The fluorescence is captured using a specialized CCD cooling camera, and the images are analyzed for variations in wavelengths/color that indicate which molecules are present.

The mote network will be similar to that previously developed by the PODS project researchers at the University of Hawaii. Dr. Edo Biagioni, PODS researcher and Computer Science Professor, provided the necessary guidance on possible network technologies, alternative power consumption issues, and legal permission regarding the use of Kilauea as the test site.

DESIGN

Many aspects of engineering and biology go hand-in-hand. This paper will have an overview of bio-detection and mote networking, but its primary focus will be on the design of the protective casing and power supply, which have been the author’s main tasks in this project this semester.

Bio-detectors

Lipid membranes have been determined to be the best indicator of life. DNA, while also an indicator of life, may not be the same on Mars as it is on Earth. However, the lipid membranes that surround and protect the inner cell seem to be common among all living organisms. To detect these lipid membranes, chemical gels will be used. The chemical gels will be released beneath the topsoil where they will bond with any lipid membranes and cause a chemical reaction that produces a fluorescent glow. Photodiode arrays will detect the specific wavelength(s) emitted. From the data collected by the photodiode arrays, it can be determined whether lipid membranes exist in the soil or not.

Networking/Communications

Jennic’s IEEE802.15.4 evaluation kit will be used to implement the wireless sensor network. Data will be sent and received via 2.4GHz IEEE802.15.4 compliant radio waves. The star networking topology will be used such that each mote is wirelessly connected to a central

100

mote that will send the desired data to the main computer for data collection and analysis. Each mote is capable of being programmed so that data can be sent at certain intervals throughout the day, rather than continuously. By collecting data at discrete times, each mote saves power by reducing power consumption during times of inactivity, thus extending the life of the system.

Casing

A cone shaped casing will enclose and protect the circuitry of each mote from the extreme environmental conditions it may face (refer to Figure 1). This conical shape has been chosen for several reasons:

• The circular base provides a strong foundation by eliminating stress points (i.e. any vertices at the base) that may cause the casing to crack upon deployment and consequently expose the circuitry to dust.

• The heavier bottom increases the probability that the mote will land and remain upright

for its duration of use.

• The angled sides reduce dust build up, allowing maximal sun absorption by the solar panels, and providing the ideal location for the antenna (at the top of the cone) where dust collection is least, keeping the communications open for the longest period of time.

Additionally, a hollow

spike, central in the casing, will provide a means of penetrating the soil to release the chemical dyes it holds. Organisms on the direct surface of Mars are more difficult to detect as the high UV radiation on the surface destroys or greatly alters genetic material, decreasing the chances that anything will be found. Therefore, by releasing the bio-chemicals below the immediate surface, the data collected would be a more accurate and in-depth look into whether life exists on Mars.

Ideally, the casing would be constructed using similar materials used for the Mars Exploration Rover; an aluminum honeycomb structure sandwiched between graphite-epoxy face sheets covered by an additional layer of phenolic honeycomb filled with ablator, a material that dissipates heat by atmospheric friction. However, due to budget limitations, off-the-shelf materials will be used instead. Unable to find an aluminum cone for a reasonable price, a traffic cone will suffice as the conical structure that will house the mote components. A pipe will be affixed as the spike that will contain the chemical dyes. The spike will be attached to the cone using epoxy glue. Solar panels will also be mounted to the cone using epoxy.

Figure 1: Cross-sectional diagram of casing showing central spike and outer solar panels.

101

Power Supply

To power the mote efficiently and for the longest period of time, it was necessary that a renewable energy source be incorporated to recharge the batteries. Solar panels mounted on the conical casing were the best option. A circuit consisting of solar cells, rechargeable NiMH batteries, and a Schottky diode will be used (refer to Figure 2).

Flexible thin film solar modules have been chosen because of its lightweight, paper thin, and durable qualities. The flexibility of the solar panels is important since it will be wrapped

around a conical casing. These solar modules have been specifically developed to recharge AA, AAA, plus 6 and 12 volt batteries.

The solar modules purchased operate at a voltage of 6.0 V and 100 mA. The size in inches (L x W x T) of each solar module is 4.5 x 5.9 x 0.01. Based on these specifications, the solar cell’s energy conversion efficiency (i.e. the power converted from absorbed light to electrical energy) can be calculated: Let η = the energy conversion efficiency Let PM = the maximum power point Let E = the input light irradiance Let AC = the surface area of the solar cells Let η = PM / (E x AC)

Substituting values into the equations for PM, AC, and η,

PM = (6 V)(0.1 A) = 0.6 W E = 589.2 W/m2 (taken from Mars data sheet) AC = (4.5 inches)(5.9 inches) = (0.1143 m)(0.16968 m) = 0.017129 m2 η = 0.6 / [(589.2 W/m2)( 0.017129 m2) = 0.059451 ≈ 5.95% The solar modules convert approximately 5.95% of the absorbed sunlight into usable energy.

Referring to the datasheet for the devices purchased, the maximum value used for the voltage and current are 3.6 V and 50 mA, respectively. The maximum power consumed would be P = (3.6 V)(50 mA) = 0.18 W, or 1.67% of a 6 V, 1800 mA battery. Therefore, each solar panel module would be sufficient in charging five batteries, or one battery pack.

The batteries used will be 6 V, 1800 mAh, nickel medal hydride (NiMH) rechargeable battery packs (each pack contains 5 AA batteries connected in series). Compared with the similar NiCd battery, NiMH batteries are more environmentally friendly as its anode is made from a hydrogen-absorbing alloy instead of cadmium. This reinforces one of this project’s goals, which was to create motes that are least invasive to the environment. Additional benefits in using NiMH batteries instead of rechargeable NiCd batteries are: lower memory effect so batteries do not require full discharge before recharge and can hold more charge, a higher internal resistance best suited for devices that do not require a lot of power, and long term maintenance by low duty cycle pulses of high current rather than continuous low current (which

Figure 2: Circuit diagram of power supply.

102

is ideal since the solar panels will be unable to provide a continuous low current to the batteries at night).

A Schottky diode will prevent the batteries from discharging through the solar cells at night. Schottky diodes are especially useful for the discharge protection of solar cells because of its low forward-voltage drop and quick switching.

CONCLUSION

Based on the design developed, a prototype mote network will be built to demonstrate the effectiveness and feasibility of using such technologies as an alternative means of detecting life in extreme environments such as Mars (i.e. compared to life detecting rovers). If initial tests in Hawaii prove successful, further tests in Chile’s Atacama Desert, a site that has been used by other NASA researches as a “Mars analog environment,” will be conducted. The data collected will be used to determine whether life exists or could exist, thus, opening up the options for further space exploration.

ACKNOWLEDGMENTS The author would like to thank:

• Mentor: Dr. Kim Binsted • Group members: Mary Liang, Joshua Irvine, and Tiffany Iiga • Sponsors: the Hawaii Astrobiology Institute, and the NASA Space Grant Consortium • References: Dale Anderson, Edo Biagioni, Andrew Boal, Victoria Hamilton, Daniel

Jenkins, Chris McKay, Wei Wen Su, and Alan Waggoner • Friends: Brian Chee, Johnson Hung, and Andrew Yasui

All of these individuals have contributed immensely to this project and it is with much appreciation that the author would like to acknowledge and thank them.

103

REFERENCES 2003 Strategic Plan 2003, National Aeronautics and Space Administration, USA. Anderson, Dale. SETI Institute Principal Investigator. [Personal E-mail] 02 February 2006. Biagioni, Edo. Associate Professor, University of Hawaii Department of Information and

Computer Sciences. [Interview] 11 January 2006. Boal, Andrew. Astrobiology Postdoctoral Research Fellow, University of Hawaii Manoa.

[Interview] April 6, 2006; [Personal Email] 10 April 2006. Hamilton, Victoria. Assistant Professor, Hawaii Institute of Geophysics and Planetology.

[Interview – Teleconference] 9 February 2006. Hooper, Joseph. “Is This the Machine That Will Finally Find Life On Mars?”. Popular Science.

January 2006. Jenkins, Daniel. Assistant Professor, University of Hawaii Department of Molecular Biosciences

and Bioengineering. [Interview] 24 January 2006. Mars Exploration Rover [On-line] 2006. http://en.wikipedia.org/wiki/Mars_Exploration_Rover,

May 2006. McKay, Chris. Planetary Scientist, Space Science Division of NASA Ames. [Personal E-mail]

06 February 2006. Smartdust [On-line] 2005. http://en.wikipedia.org/wiki/Smartdust, 08 December 2005. Su, Wei Wen Winston. Professor, University of Hawaii Department of Molecular Biosciences

and Bioengineering. [Personal E-mail] 01 April 2006. Waggoner, Alan. Director of the Molecular Biosensor and Imaging Center, Professor of

Biological Science and Biomedical Engineering, Carnegie Mellon University. [Personal E-mail], 28 March 2006.

Williams, David R., Mars Fact Sheet [On-line] 2006.

http://nssdc.gsfc.nasa.gov/planetary/factsheet/marsfact.html, 01 September 2004.

104

LEONIDAS SATELLITE CONCEPT STUDY REPORT

Team Lead: Lloyd French

Team Support: Gindi French, David Hampton

LSCS TEAM: Dennis Dugay

Aukai Kent Zachary Lee-Ho

Matthew Patterson

College of Engineering

University of Hawaii at Manoa Honolulu, HI 96822

105

ABSTRACT

Microsatellites provide a cheap and efficient way to perform experiments, collect data and provide space validation for science technology developments. Hawaii has the capability to complete an entire low Earth orbiting satellite mission which gives it a tremendous advantage in microsatellite market. Our satellite concept report is an outline of how we would construct a satellite with “plug and play” capabilities to allow easy integration of various payloads. We have listed the various components for each subsystem and specified how each will interact with one another. The concept report will serve as a foundation for writing future proposals.

BACKGROUND

Extreme high cost of manufacturing and launching large satellites has increased need for micro and nanosatellites to take this role. A typical large satellite can have a mass over 10,000 kg when you include the fuel and cost over 300 million dollars to manufacture and launch. However, microsatellites way between 10 and 100 kg and meet the high performance demands but cost approximately 1% of what it takes to build a large satellite. The State of Hawaii is unique in the fact that it is one of the few states with the capability to complete an entire low Earth orbiting satellite mission. Along with an abundance of experience in the engineering and science disciplines necessary to complete Phase A proposals and mission designs, Hawaii also has the facilities to manufacture, integrate and test space flight hardware. Through the Pacific Missile Research Facility (PMRF) on the island of Kauai, Hawaii has the capability to launch small satellites into polar orbit. Also, unlike land locked states, Hawaii does not have the nominal risks of failing debris from stages one and two. Finally, Hawaii has antennas capable of receiving and sending commands and controls to satellites in orbit. All these capabilities give Hawaii a tremendous advantage in the microsatellite market.

In 2002 Dr. Luke Flynn, director of the Hawaii Space Grant Consortium (HSGC), Dr. Wayne Shiroma and Dr. Carlos Combria, professors of the college of Engineering, founded the Low Earth Orbit NanoSatellite Integrated Distributed Alert System (LEONIDAS). LEONIDAS originated as an extension of the CubeSat program created by Dr. Shiroma, in response to the Department of Defense’s (DOD) increased interest in developing smaller satellites. LEONIDAS eventually evolved to focus on developing satellites for rapid deployment and repetitive surveillance of a given area. Since the development of LEONIDAS, the University of Hawaii Engineering Department has constructed three student-developed cubesats, one of which is waiting in Russia to be integrated and launched with other universities’ cubesats. University of Hawaii gained further experience in the space technology field by funding two engineering students from HSGC, to participate in the Magnetic Field Investigation of Mars by Integrating Consortia (MIMIC) mission proposal hosted at Jet Propulsion Laboratory’s (JPL).

Recently Professor Lloyd French, a System Architect for LEONIDAS, sought to create a team of undergraduate students to design and launch a microsatellite. The team is comprised of students from the Department of Engineering and Geophysics & Planetary Sciences that includes Native Hawaiian and local engineering students. The microsatellite will demonstrate Hawaii’s capability to complete an entire spacecraft mission and serve as a testbed for experiments and science technology developments.

106

OBJECTIVE The LEONIDAS Mission Concept Study Team (LMCST) objectives are to research and understand the basic concepts for spacecraft design. We will research the various satellite components, payloads, past successful launched buses and how the subsystems will integrate and function with one another. The concept study will be colossal in advancing our education as engineering students through the knowledge and experience we will gain while working on the LMCST. Our report will demonstrate Hawaii’s capability to complete a successful spacecraft mission, and from our concept study we will develop future proposals.

GOALS Our primary goal is to submit a proposal to be entered into the Air Force’s nanosatellite competition. Given funding we wish to start developing the prototype for our satellite. Ultimately, we want to build, launch and operate a fully functional satellite carrying GPS experiment, an antenna experiment, software experiment and an imager for global reconnaissance. A successful mission would place the University of Hawaii’s College Engineering program amongst the elite engineering programs in the nation. This would have major implications in enhancing the technology and research at the university. It would also be beneficial in the recruitment of finest professors and the best and brightest students of Hawaii. The mission’s success would be a major stepping stone for improving the economic development of Hawaii.

SYSTEMS ENGINEERING

The baseline design for the LMCST satellite concept was intended to conduct four experiments. The first experiment will conduct remote sensing using UV imaging. The last two experiments will be test flown to demonstrate their effectiveness for future missions. They are an Active Antenna by Dr. Wayne Shiroma’s CubeSat team and a GPS unit provided by SSTL. A fourth experiment involves testing a spacecraft housekeeping software provided by JPL. These experiments will operate in a sun synchronous orbit with a period of 96 minutes at about 350 km. Our satellite will orbit the earth approximately 15 times a day in which only three will be adequate to perform experiments, operate the imager or transfer data.

There are eight different modes for the spacecraft: Camera Experiment, Active Antenna Experiment, SSTL Experiment, JPL Experiment, Communications, Maneuver, Power Save, and Safe Mode. During the Camera Mode the spacecraft will image a specified area. In the Active Antenna Experiment Mode we will attempt to communicate to the ground station by running Active Antenna and the UHF unit to demonstrate its functionality. The SSTL GPS unit will run during SSTL Experiment Mode were it will attain and collect knowledge on our spacecraft’s polar orbit. The JPL Experiment Mode will take place after primary mission is complete is a software that monitors the health of the spacecraft. The spacecraft’s Communications Mode utilizes the S-band frequency to transmit and receive data to and from the ground station. During the Maneuver Mode of the spacecraft uses the ACS system to adjust its orientation. While in darkness, the spacecraft will conserve energy by entering a Power Save Mode where all systems,

107

except for C&DH will function in standby. In an emergency, the Safe Mode will be utilized where by the spacecraft shuts down all systems excluding the secondary communications and the C&DH systems.

By incorporating these modes into an orbital period we were able to construct the sample operations cycle in Table 1. There will be four different operation cycles for our mission. The operations cycle is used to create a power profile for the spacecraft, which is used to size the power system. The power profile for communications and payloads is shown Figure 1 & 2.

Time (min) Mode Power (W) Time (min) Mode Power (W)

0 SSTL Experiment 42.36 0 Active Experiment 44.486 SSTL Experiment 42.36 6 Active Experiment 44.48

12 Power Save 14.8 12 Power Save 14.818 Power Save 14.8 18 Power Save 14.824 Power Save 14.8 24 Power Save 14.830 Power Save 14.8 30 Power Save 14.836 Power Save 14.8 36 Power Save 14.842 Power Save 14.8 42 Power Save 14.848 Power Save 14.8 48 Power Save 14.854 Power Save 14.8 54 Power Save 14.860 Power Save 14.8 60 Power Save 14.866 Power Save 14.8 66 Power Save 14.872 Power Save 14.8 72 Power Save 14.878 Power Save 14.8 78 Power Save 14.884 Maneuver 41.36 84 Maneuver 41.3690 SSTL Experiment 42.36 90 Active Experiment 44.4896 SSTL Experiment 42.36 96 Active Experiment 44.48

Time (min) Mode Power (W) Time (min) Mode Power (W)0 Communications 50.36 0 Camera Experiment 47.366 Communications 50.36 6 Camera Experiment 47.36

12 Power Save 14.8 12 Power Save 14.818 Power Save 14.8 18 Power Save 14.824 Power Save 14.8 24 Power Save 14.830 Power Save 14.8 30 Power Save 14.836 Power Save 14.8 36 Power Save 14.842 Power Save 14.8 42 Power Save 14.848 Power Save 14.8 48 Power Save 14.854 Power Save 14.8 54 Power Save 14.860 Power Save 14.8 60 Power Save 14.866 Power Save 14.8 66 Power Save 14.872 Power Save 14.8 72 Power Save 14.878 Power Save 14.8 78 Power Save 14.884 Maneuver 41.36 84 Maneuver 41.3690 Communications 50.36 90 Camera Experiment 47.3696 Communications 50.36 96 Camera Experiment 47.36

Operation Cycles

Table 1: Operation cycles of satellite while in orbit.

108

Power Profile for Communication System

0

10

20

30

40

50

60

Com

mun

icat

ions

Com

mun

icat

ions

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Pow

er S

ave

Man

euve

r

Com

mun

icat

ions

Com

mun

icat

ions

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96

Mode Time Interval (minutes)

Power Profile for Payloads

0

5

10

15

20

25

30

35

40

45

50Payload

Payload

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Power S

ave

Maneuver

Payload

Payload

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96

Mode Time Intervals (minutes)

CAMERAACTIVE ANTENNASSTL

Figure 1: Power profile for Communication System

Figure 2: Power profile for Payloads

109

Components for each subsystem are chosen to fit within the specifications of the above power profile. In the flow chart in Figure 4 is an outline of the six subsystems integrated in our spacecraft, they are:

• Payload • Command & Data Handling • Attitude and Control • Communications • Power • Thermal

The components for each subsystem are grouped together and color coordinated for easy readability. A dash line shows power distribution, and the best approximation of power each component needs is listed inside each box. There are four major interfaces that re used to transfer data throughout the spacecraft they are IEEE 1394 and RS232/42/485. The transfer of data is represented by a solid black line and flows in the direction the arrows are pointing. Those subsystems and payloads that must be switched on and off at designated times during the polar orbit are connect to the control board by a solid red line. Each component’s specifications and why they were chosen is explained in further detail in the following subsystems report.

University of HawaiiActive Antenna

Mass 26 gDim. 50 x 70 mm

Power 0.54 W

QImagingUV CameraMass 595 g

Dim. 76 x 64 x 120 mmPower 6W

Data Rate 40 MB/s

Surrey Satellites Surrey GPS Experiment

Mass 1150 gDim. Power 6.3 W

Diamond SystemsProcessor/ Data Acquisition

Mass 79 gDim. 90 x 96 mm

Power 5WPC-104

Diamond SystemsMemory module

Mass 35 gDim. 90 x 96 mm

Capacity 4 GBPC-104

YaesuUHF Transceiver

Mass 85 gDim. 94 x 56 x 13.6 mm

Power 0.88 W (trans) 0.15 w (rec)

Aero AstroTransmitter

Mass <200 gDim. 89 x 51 x 26 mm

Power 8WDownlink 124 Š 256 kbps

Aero AstroReceiver

Mass <200 gDim. 89 x 51 x 26 mm

Power 0.5 W (stdby), 1 W (oper)Uplink 1, 2, or 10 kbps

AntcomAntenna

Mass 270 gDim. 45 x 34 x 122 mm

Power Handling 5 W

SpectrolabSolar Panels

Power Output 350 W/m 2

Mass 1760 g/m 2

Battery ModuleDim. 180 x 192 x 210 mm

Power 90 WphrMass 2948 g

PC-104

Tri-MPower Supply Module

Mass 186 gDim. 90 x 96 x 14 mm

Power output 3.3, 5, 12, -12Input Power 6- 40 V DC

PC-104

Tri-MGPS Antenna

Mass 65 gDim. 48 x 15 x 58 mm

Power 0W

University of HawaiiThrusters

MassDim. mmPower W

HoneywellReaction Wheels

Mass 5200 gDiameter >130 mm

Height > 54 mmPower < 24 W

XsensIMU

Mass 50 gDim. 58 x 58 x 22 mm

Power .36W

Honeywell7 x Temperature Sensor

Platinum RTD Temp. Range -200

PacCommTNC (3)

Mass 160 gDim. 25 x 63 x 83 mm

Power 0W

AMPLTDInterface board

Mass 85 gDim. 90 x 96 mm

Power 3WPC-104

Diamond SystemsControl Board

Mass 85 gDim. 90 x 96 mm

Power 1.45 WPC-104

Diamond SystemsInterface board

Mass 79 gDim. 90 x 96 mm

Power 0.4 WPC-104

Power

SunSpaceGyroscopeMass 439 g

Dim. 118 x 118 x 115 mmPower < 2 W

ParvusGPS ModuleMass 85 g

Dim. 90 x 96 mmPower 3.2 W

Control Board Switch

RS-232

RS-232

RS-232

RS-485

RS-422

RS-422

RS-232

S-Band

RS-232

UH

F

IEEE 1394 U

HF

IEEE 1394

S-BandS-Band

RS-232

Data

C&DHADSCOMMPOWERTHERMALPAYLOAD

RS-422

Figure 3: Flowchart of satellite’s subsystems

110

PAYLOADS

Payloads are merely the instruments, devices, imagers or experiments the spacecraft carries and houses to perform the given tasks for the desire mission and is the main reason for developing the spacecraft initially. Our philosophy when developing our satellite concept design was to create a bus that possesses the capabilities to easily integrate and deploy various payloads. We hope to provide a means to achieve space validation for new technologies and act as a testbed for low earth orbit experiments.

To establish a baseline of what components are needed in order to develop our satellite we have chosen three payloads that mark the furthest capability that the spacecraft can deliver. To scope the capabilities of our satellite we have chosen an imager, a low earth orbit experiment and a science technology development from the various agencies interest in our satellite concept design. Ultra-Violet & Visible Camera

The primary payload on our satellite will be an ultra violet camera that will have the ability to take pictures in both visible and ultra-violet light. Pictures taken by the UV camera will provide proof of our launch and can be used as a means to monitor the ecosystem of Hawaii. Picture taken in visible light could help environmental and economic development.

Requirements • IEEE 1394 FireWire Interface • Power Supply 8-24 V • Sustain Data Rate 40 MB/s

We will be using the QICAM-UV by QImaging, a digital camera that has high resolution UV and visible range with scientific and industrial imaging applications. QICAM-UV has a spectral

range that extends to 200 nm in the UV region and provides a resolution of 1.4 million pixels in a 12-bit digital output. With a high-speed readout it produces linear image data at a maximum frame rate of 205 fps.

Specifications (UV-Camera) Dimensions 76 x 64 x 120 mm

Operating Temperature 0ºC/+35ºC Mass 595 g

Power Consumption 6 Watts Exposure/Integration Control 12 µs to 17.9 min in 1 µs increments

Optical Interface ½”, C-Mount optical format Light Sensitive Pixels 1.4 million; 1392 x1040

Pixel Size 4.65 µm x 4.65 µm Doppler Velocity 0.5 m/s

Figure 4: QICAM-UV

Table 2: Specifications for QICAM-UV

111

Active Antenna

Dr. Wayne Shiroma and The University of Hawaii CubeSat team developed the Active Antenna known as a grid oscillator. The grid oscillator was developed for application with The University of Hawaii’s CubeSat and as an alternative to the usual low frequency range UHF/VHF.

Requirements • Power Supply 1.5 V • 4 mm buffer beyond the housing wall

The grid oscillator transmits at higher frequencies and

has an efficient power-combining scheme with a compact design. Unlike conventional wire or strip antennas the grid oscillator does not need to be deployed nor does it require additional circuitry. The grid oscillator’s built in redundancy makes it tolerant to single point failure, which decreases the

chance for component failure. If the Active Antenna is successful it will help lay the groundwork for future data-intensive CubeSat missions.

Specifications (Active Antenna) Dimensions 50 x 70 x 12.7 mm

Mass 26 g Operating Frequency 5.76 GHz Power Consumption 0.54 Watts

Transmitted Power 20.33 dBm Surrey Satellite Technology Ltd. Space GPS Receiver

Surrey Satellite Technology Ltd. has interest in testing their technology on our satellite. Surrey will aid us in integrating their GPS receiver to our bus.

Requirements • Power Supply 18-38 V • RS422 Interface • Minimum three antennas to determine altitude

Surrey GPS receivers are known for providing GPS standard time,

position and velocity in a compact unit. From 24 hours of data Surrey’s GPS receivers have the ability to deliver onboard orbit knowledge to within several meters. Surrey’s SGR-05/10/20 GPS receivers each decode and receive L-Band signals, and have the ability to calculate the position within 10 meters. The code is stored in Flash memory which enables the receiver to boot rapidly and gives it the ability to upgrade while in orbit. In addition, Surrey GPS receivers have a separate TTC node that gives it telemetry and telecommand from the primary processors.

Table 3: Specifications for Active Antenna

Figure 5: CubeSat Team’s Active Antenna

Figure 6: SSTL GPS Receiver

112

Specifications (SSTL GPS Receiver)

Dimensions 160 x 160 x 50 mm or 295 x 160 x 35 mm Operating Temperature -20ºC/+50ºC

Mass 950 g (Unit), 50 g (antenna) Power Consumption 6.3 Watts (28 V) Orbital Position (3-D) 10 - 20 m Orbital Velocity (3-D) 0.15 - 0.25 m/s

Attitude Determination 0.5º - 1º Pseudorange 0.9 m

Doppler Velocity 0.5 m/s

COMMAND & DATA HANDLING (C&DH)

The Command & Data Handling subsystem collects data and telemetry of the satellite and manipulates data sets from payload. It runs and stores all necessary software for the spacecraft and controls all functions carried out by each subsystems. The C&DH implements all commands and direction received by the ground station and relays the data to the correct subsystem. It directs the modes of the spacecraft while it orbits the Earth and ensures each one is operating at the proper time. Each board in our C&DH subsystem will use the PC-104 form factor, which will make assembly of the subsystem easier and thus eliminate possible error created through integration of different buses. Processor Board

The processor board is the brain of the entire spacecraft. It processes all data and runs all software for each subsystem. Limited by the total mass and power of spacecraft, it was important to have a small, low power consumption processor. We also want our board to use the PC-104 bus because it will simplify the assembly when integrating each subsystem.

Requirements • Mass < 100 g • Power Consumption < 5 W • PC-104 Bus Interface • Processing Speed > 100 MHz • Be able to withstand temperatures -40ºC to 85ºC

Our spacecraft uses a Prometheus Z32-E-ST, manufactured by Diamond Systems Corporation, a two in one processor and data acquisition board to control each task. The built in data acquisition saves space which we are limited on and prevents us from needing to incorporate an additional board. Prometheus processor uses a ZFx86 microprocessor chip, by ZF Micro Solutions Company, with the ZF FailSafe System that has the

Table 4: Specifications for SSTL GPS Receiver

Figure 7: Prometheus Processing Board

113

ability to recover from software crashes or operating system corruption without human intervention. The Prometheus is a low power cost processor with industrial temperature rating. Besides ensuring that all subsystems are running properly the processor will process pictures taken from the UV-camera and run the software for Dr. Shiroma’s and Surrey’s experiment. The Prometheus’ speed of 100MHz with only 5 watts of power is plenty. The Diamond Systems Corp. offers the option of adding a flashdisk that will allow us to store up to 4GB of data between downlinks.

Specifications (Prometheus) Processor ZF Micro

ZFx86 Speed 100 MHz

Power Consumption 5 Watts Serial 4 RS-232

Ethernet 10/100 USB 2 X 1.1

# Inputs 16 SE , 8 D/I # Outputs 4

# Digital I/O 24 I/O Operating Temperature -40ºC /+85ºC

Dimensions 90 X 96 mm Interface Boards

The interface board provides a way for all the subsystems to communicate to the processor board. It acts as translator of data received from the Payloads, Attitude Control and Communication System. The interface boards are key elements in our mission concept which will demonstrate the ability to produce a microsatellite that has “plug and play” capabilities i.e., fully operational without the need build or design special interfaces to make components compatible. The “plug and play” capability will allow us to maneuver components around in the spacecraft to accommodate payloads and experiments. Requirements • Mass < 100 g • Power Consumption < 3 W • PC-104 Bus Interface • IEEE 1394 FireWire with data rate of 400 MB/sec • 4 < RS-232 Ports • 2 < RS-422 Ports • 2 < RS-485 Ports • Be able to withstand temperatures -40ºC to 85ºC

Figure 8: Flashdisk memory

Figure 9: FireSpeed2000 Interface Board

Table 5: Specifications for Prometheus Processing Board

114

The C&DH subsystem contains two interface boards the Emerald-MM and the FireSpeed2000. The interface boards consist of IEEE 1394 and RS-232/422/485 ports, which makes integration of the UV-Camera and the other subsystems possible. The Surrey Satellite Experiment has a TBD interface, but we have approximately 9 watts of power and 5kg to accommodate any necessary boards that need to be integrate for the experiment.

Control Board

The control board acts as the operations director it ensures that only the proper subsystems are running and those not needed are off. The control board is more specifically a

counter/timer and digital input/output module. Requirements • Mass < 100 g • Power Consumption < 2 W • PC-104 Bus Interface • 7 < Counters • Be able to withstand temperatures -40ºC to 85ºC

We will use a Quartz-MM-10 (QMM-10) to control the

operations of the spacecraft. The QMM-10 has 10 counters each 16 bits wide, which is enough for each payload and every subsystem. It is also made by Diamond Systems Corporations so integration with the other C&DH boards should be no problem.

Specifications (FireSpeed 2000) Interfaces IEEE 1394 FireWire

Transfer rate 100/200/400 MBits/sec PC-104 Capable Yes

Power Consumption 3 Watts Operating Temperature -40ºC /+85ºC

Dimensions 90 X 96 mm

Specifications (Emerald-MM) Interfaces RS-232/422/485

PC-104 Capable Yes Power Consumption 0.4 Watts

# RS-232 ports 4 # RS-422 ports 2 # RS-485 ports 2

Operating Temperature -40ºC /+85ºC Dimensions 90 X 96 mm

Figure 11: Quartz-MM-10 control board

Figure 10: Emerald-MM Interface Board

Table 6: Specifications for Emerald-MM Interface Board Table 7: Specifications for FireSpeed 2000 Interface Board

115

Specifications (Quartz-MM-10) # Inputs/ Outputs 8 in, 8 out

PC-104 Capable Yes Power Consumption 1.45 Watts

#Counters 10 Resolution 16 bits

Max clock input rate 20 MHz Operating Temperature -40ºC /+85ºC

Dimensions 90 X 96 mm

ATTITUDE CONTROL SYSTEM (ACS)

The Attitude Control System determines and controls the spacecraft’s location in space and its orientation relative to the Earth. The ACS is vital for placing the spacecraft in the proper position to carry out experiments, pointing for taking pictures and downlinking and uplinking data.

Global Positioning System (GPS) Receiver

The GPS subsystem allows us to compute the spacecraft’s position and velocity. Like the C&DH subsystem we want the GPS module to have the PC-104 architecture to allow for easy integration with C&DH subsystem. Requirements • Mass < 100 g • Power Consumption < 5 W • PC-104 Bus Interface • Be able to withstand temperatures -40 ºC to 85ºC

We will use the COMM-1288 GPS module by Parvus for our spacecraft. The COMM-1288

integrates on a single PC-104 board, a high-speed Triband 900/1800/1900MHz GSM/GPRS modem and a low power 12-channel parallel tracking GPS receiver.

Specifications (Comm-1288) GPS Receiver iTrax02

Power Consumption 1.1 Watts (idle) 3.2 Watts (peak)

# Serial Ports 4 x RS-232 Mass 85 g

Operating Temperature -40 ºC/+85 ºC Dimensions 90 x 96 x 15 mm

Figure 12: COMM-1288 GPS Receiver

Table 8: Specifications for Quartz-MM-10 control board

Table 9: Specifications for COMM-1288 GPS Receiver

116

GPS Antenna

We will use the Big Brother GPS Antenna by Tri-M Systems and Engineering. The GPS Antenna allows us to amplify the GPS signal to the GPS receiver. We chose the Big Brother antenna for its durability and for its built in low noise amplifier.

Orientation System

The GPS system will give the spacecraft’s location in space, but the orientation to Earth will be determined using the Inertial Measurement Unit (IMU) and a gyroscope. The IMU tracks rapidly changing orientations in 3D and measures the directions of gravity and magnetic north to provide a stable reference, and the gyroscope will provide additional 3-axis inertial measurements. We will use an IMU by Xsens and a gyroscope by SunSpace. The Sunspace Gyroscope has flight heritage on TUBSAT.

Specifications (GPS Antenna) Center Frequency 1575.42 MHz

Noise Figure 2.0 max Bandwidth 2MHz min.

Output Impedance 50 ohm Mass 65 g

Operating Temperature -40 ºC/+85 ºC Dimensions 48 X 15 x 58 mm

Specifications (IMU) Dynamic Range All angles in 3D

Angular Resolution 0.05 deg Power Consumption 0.36 Watts

Digital Interface RS-232 Mass 50 g

Operating Temperature 0 ºC/+55 ºC Dimensions 58 x 58 x 22 mm

Specifications (Gyroscope) Measurement Range 80 deg/s Power Consumption < 2 Watts

Data Interfaces RS-232/485 Mass 439 g

Dimensions 99 x 117 x 31 mm

Figure 14: IMU Figure 15: Gyroscope

Table 10: Specifications for GPS Antenna Figure 13: GPS Antenna

Table 12: Specifications for Gyroscope Table 11: Specifications for IMU

117

Reaction Wheels

Reaction wheels allow us to adjust the spacecraft’s orientation and inertial rates. The spacecraft will have three reaction wheels one for each axis of motion. Requirements • Mass < 2000 g • Power Consumption < 7 W • Data Interface RS-232/422/485 • Be able to withstand temperatures -25 ºC to 55ºC

We chose to use Honeywell’s miniature reaction wheels to control our spacecraft’s momentum. The Honeywell gyroscope is built based on commercial and military aircraft gyroscope technology. It has a compact design with high momentum-to-mass efficiency and uses a DC brushless-spin motor.

Specifications (Reaction Wheels) Angular Momentum 0.2 to 1.0 Nms

Wheel Torque Output > 28 mNm Wheel Speed Range + 9000 rpm

Data Interfaces RS-422 Mass 1300 g

Operating Temperature -25 ºC/+60 ºC Dimensions 130 mm (diameter)

54 mm (height) Power Consumption < 6 Watts

Figure 16: Reaction Wheel

Table 13: Specifications for Reaction Wheel

118

COMMUNICATION SYSTEM (COMM)

The COMM subsystem transmits and receives all data and commands between the spacecraft and the receiving ground station. We will use S-Band frequency to send information, in part because X-Band, though faster, is a frequency that is over used, therefore, making it difficult to communicate to our spacecraft due to “cross talking”. Receiver (S-Band)

The receiver’s main objective is to collect commands and directions sent from the command station on earth and relay them to the processor. Requirements • Mass < 500 g • Power Consumption < 8 W • Uses Interfaces RS-232/422/485 • Capable to receive S-Band frequency • Uplink rate > 10 kbps

For our spacecraft we would use the S-Band receiver by AeroAstro Company. The AeroAstro receiver is one of the lowest power cost receiver; with only one watt of power, it uplinks data at a rate of 10 kbps. The receiver module is compact in size at 89 mm x 51 mm x 25 mm, making it no longer than a PC-104 board.

Transmitter (S-Band)

The transmitter relays all data received from the payloads and information concerning the status of each subsystem to the command station. The transmitter can also act as a Doppler range finder to locate the spacecraft in orbit by measuring the time a signal is sent out and returns back. Requirements • Mass < 500 g • Power Consumption < 35 W • Uses Interfaces RS-232/422/485 • Capable to receive S-Band frequency • Uplink rate > 125 kbps

Specifications (Receiver) Receive Frequency 1760 – 1840 MHz

Interface RS-422 Power Consumption 0.5 Watts (stdby)

1 Watts (operational) Input Data Rate 10 kbps

Operating Temperature -30 ºC/+60 ºC Dimensions 89 X 51 x 25 mm

Mass < 200 g Figure 17: Inside an AeroAstro communication module Table 14: Specifications for Receiver

119

Table 16: Specifications for S-Band Antenna

We will also being using an AeroAstro S-Band transmitter for our spacecraft. The AeroAstro transmitters have been flight-proven and were used in the Canadian Space Agency’s MOST mission in which it is still fully functional. Once again the AeroAstro module was the lowest power cost module we discovered. It has the ability to downlink at 124-256 kbps with only 8 W DC. Specifications (Transmitter)

Transmit Frequency 2200 – 2300 MHz Interface RS-422

Power Consumption 8 Watts Output data rate 125 – 256 kbps

Operating Temp. -30 ºC/+60 ºC Dimensions 89 X 51 x 25 mm

Mass < 200 g Antenna (S-Band)

The S-Band antenna acts as a beacon for the transmitter and receiver. Data is received and release through the antenna. On our spacecraft we will have two S-band antennas for the transmitter and receiver. When searching for an S-Band antenna our main concern was to find

the smallest antenna possible in order attain the most surface area possible for the solar panels. Requirements • Mass < 500 g • Capable to receive S-Band frequency • Operational altitude > 4.83 km • Dimensions < 45 x 140 mm

We will use Antcom’s S-Band antenna for its small size and its durability. It has the ability to withstand altitudes up to 21.3 km and vibrations up to10 G’s.

Specifications (S-Band Antenna) Frequency S-Band

Power Handling 5 Watts Vibration 10 G's

Altitude 21.336 km Material 6061-T6 Al alloy base

Operating Temperature -55 ºC/+85 ºC Mass 270 g

Dimensions 44.5 x 133.4 x 121.2 mm

Figure 18: AeroAstro Receiver/Transmitter

Figure 19: S-Band Antenna

Table 15: Specifications for Transmitter

120

UHF Transceiver

The UHF Transceiver will collect and release data during Active Antenna experiment. It will also act as a backup for the S-Band transmitter and receiver if they should happen to fail. Requirements • Mass < 500 g • Capable to receive UHF frequency • Uses Interfaces RS-232/422/485

We will use the same transceiver Dr. Shiroma’s team picked for their cubesat which is an Alinco DJ-CST transceiver built by Yaesu Company. Dr. Shiroma’s team chose the transceiver due to its relatively low power consumption, price and compact size.

Specifications (UHF Transceiver) Dimensions 94 x 56 x 13.6 mm

Operating Temperature -10 ºC/+60ºC Power Consumption 0.15 Watts (rec)

0.88 Watts (trans) Mass 85 g

Terminal Node Controller (TNC)

We will use three PicoPacket TNCs by PacComm to assist the Communication modules. The TNC is a modem that runs packet firmware that allows you to send a message containing a long status report, such as listing various types of equipment, service data, and current condition.

The firmware arranges the message into “packets” each having specific length and formatted into bits. The TNC then sends the formatted data to the transmitter where it is transmitted. The receiving station will be running the same TNC protocol and knows that each packet should be a certain length. If interference causes the TNC not to receive the expected message, the receiving TNC automatically tells the transmitting TNC what it received and what it did not. The transmitting TNC re-transmits the missed packets. This occurs at the speed of light and results in perfect data transmission.

Specifications (TNC)

Power Consumption 0 Watts Data Interfaces RS-232

Mass 160 g Dimensions 25 x 63 x 83 mm

Figure 20: Yaesu Transceiver

Figure 21: TNC

Table 17: Specifications for Yaesu Transceiver

Table 18: Specifications for TNC

121

Table 19: Specifications for Spectrolab Solar Panels

POWER SYSTEM

The Power subsystem consists of solar panels and batteries that provide all the necessary power for each subsystem and payload to run properly. Solar Panels

Solar panels are vital when making an autonomous spacecraft. They are responsible for powering the subsystems and payloads and recharging the battery packs. Our main objective when choosing solar panels is to attain the most efficient and least dense solar panel. To eliminate mechanical error we have chose not to deploy the solar panels. Solar panels will be attached both magnetically and non-magnetically. Requirements • Mass < 2 kg/m2 • Power > 290 W/m2

We will use ultra triple junction (GaInP2/GaAs/Ge) solar cells, made by Spectrolab that will provide 350 W/m2 with a mass of 1.76 kg/ m2 to our spacecraft. While directly in the sun our spacecraft will have three faces fully exposed. Each face has a surface area of 0.07 m2 that will give our spacecraft a minimum 73.5 watts of power for the entire spacecraft. That gives us a margin of 10 watts from total power, computed for the entire spacecraft.

Specifications (Solar Panels) Power 350 W/m2 Mass 1.76 kg/m2

Material GaInP2/GaAS/Ge Low Earth Orbit life Expectancy 5 years

Figure 22: Spectrolab Solar Panels

122

Battery packs

Battery Packs provide a way to supply energy to the spacecraft while it is eclipsed by the Earth and store power while directly in the sun.

Requirements • Mass < 2 kg • Output voltage > 8.4 V • Operational Temperature > -40 to 65ºC

Energy on the spacecraft will be stored using Sealed Lead

Acid BAT104-SLA45 battery packs manufactured by Diamond Systems Corporation. The BAT104-SLA45 battery packs are built in PC-104 form so they are easily mounted and assembled with other

PC-104 boards. The pre-built battery pack eliminates need for assembly of our own battery module and lessens the chances for error.

There will be two battery packs each with a max output voltage of 10 V and a storage capacity of 4.5 A-Hr.

Power Supply Board

The power supply board distributes power received from battery packs and solar panels to the proper payloads and subsystems. Requirements • Mass < 200 g • Power Input range 6 to 40 V • PC-104 Bus Interface • Output Voltage range of 3 to 12 V • Operational temperatures -40 to 85ºC

Our C&DH system contains three power

supply boards that are capable of distributing power of 3.3, 5, 12, -12 V and each has the ability to receive power of 6-40 V DC.

Specifications (Battery Packs) Max output voltage 10 V

Capacity 4.5Ah Dimensions 90 x 96 x 106 mm

Mass 1474 g Temperature range -65 to 65ºC

Figure 23: BAT104-SLA45

Table 20: Specifications for BAT104-SLA45

Figure 24: Power Supply Module

123

THERMAL SYSTEM

The Thermal subsystem will monitor and regulate the temperature on each payload and subsystem. Temperature regulation is vital for the success of the mission to ensure that all subsystems and payloads stay operational. Temperature Sensors The temperature of each payload and subsystem will be monitored through temperature sensors placed on every board and payload. Requirements • Temperature Range -60 ºC to 90ºC • Accuracy < 1 ºC

We will use Honeywell’s HEL-700 Platinum RTD to provide linearity accuracy, stability and interchangeability. Resistance on HEL-700 changes linearly with temperature. The HEL-700 has a large temperature sensing range of -200 to 540 ºC.

Heaters

ThermofoilTM heaters by Minco will be controlled by the C&DH systems to ensure that none of the subsystems or payloads freezes while the Earth eclipses the spacecraft during its orbit. The ThermofoilTM heaters can safely run at wattages twice those of their wire-wound equivalents. ThermofoilTM heaters transfer heat more efficiently, over a larger surface area, than round wires.

Specifications ThermofoilTM Material Kaplan/FEP

Temperature Range -200 to 200ºC Max Resistance Density 8-70 ohms/cm²

Thickness .25 mm Mass 0.04 g/cm²

Specifications (Temperature Sensors) Sensor Type 100 Ohm Platinum RTD

Temperature Range -200 ºC to 540ºC Temperature Coefficient .00375 Ohm/Ohm/ºC

Packaging Type Radial chip, SMT axial flip chip Self Heating > 0.3 mW/ºC

Figure 26: ThermofoilTM

Table 21: Specifications for Temperature Sensor

Table 22: Specifications for ThermofoilTM

Figure 25: RTD Temperature Sensor

124

Dissipation of Heat

We will dissipate heat from the spacecraft by placing radiator panels, conductive strapping and shear panels in the spacecraft. Conductive strapping and shear panels will transfer heat from the subsystems to the radiator panels where it will then be released into space. Our orientation in orbit will also assist in protecting the subsystems thermally. Our spacecraft will orbit the Earth with its octagon side facing space, which allows us more surface area to dissipate heat and ensure that the subsystems stay cool.

SPACECRAFT BUS

The initial conceptual design of the satellite can be seen in Appendix A. The satellite is divided into four sections by shear panels, in which the components for each subsystem are mounted to. Components are mounted to shear panels to help disperse heat, minimize interface issues and eliminate the need for extra wiring.

CONCLUSION

Our next step with the completion of our satellite concept study is to finish our Pre-Phase A analysis. Afterwards we will do a cost analysis of our spacecraft structure and its components. We plan to begin construction of our prototype spacecraft in mid August and anticipate launching in 2008.

ACKNOWLEGEMENTS

LMCST would like to thank Space Grant Consortium and Dr. Luke Flynn for allowing us the opportunity to work and learn from the professors and faculty of the LEONIDAS program. We would also like to thank our mentor, Lloyd French for passing on his extensive knowledge and insight of the space technology field and life. We are truly grateful for him taking a chance on four engineering students and allowing us to be apart of his dream for microsatellite development in Hawaii. A big thank you goes to Gindi French who has played an influential role in the development of this report. We appreciate her patience and the Saturday mornings she gave up in order to sharpen our technical writing skills. We want to thank David Hampton for giving us the ability to create a visual of our satellite concept. We wish to show our gratitude to Marcia Sistoso for her patience and attentiveness to ensure we meet our deadlines and have everything we need. And to the many others that had a hand and words of encouragement during the development of this report, Thank You.

125

REFERENCES

Diamond Systems Corporation Website: (April 2006) Available [Online] http://www.diamondsystems.com/products/

Tri-M Systems and Engineering Product index Website: (April 2006) Available [Online]

http://www.tri-m.com/products/index.html Minco Product Catalog: (April 2006) Available [Online]

http://minco.com/uploadedFiles/Products/Thermofoil_Heaters/Hs202.pdf Parvus Corporation Product Info Website: Comm-1288 module. (April 2006) Available [Online]

http://parvus.com/products/MilitaryAerospace/Board-LevelOEM/COM-1288/ Honeywell International Inc. Aerospace Website: Wheel Assemblies. (April 2006) Available

[Online] http://www.honeywell.com/sites/aero/Pointing-Momentum-Control3_C80E53B46-7939-1874-4273-9D8809AFB783_HE1E6DE76-F07D-B4AA-4011-69C4BA284723.htm

Xsens Technologies Website: MTi product info sheet. (April 2006) Available [Online]

http://xsens.com/download/MTi_leaflet.pdf Q Imaging Product Info Website: QICAM UV. (April 2006) Available [Online]

https://mail.hawaii.edu/attach/QICAM-UV.pdf?sid=&mbox=space%20project&charset=escaped_unicode&uid=35&number=4&filename=QICAM-UV.pdf

Spectrolab Info Website: Solar Panels. (April 2006) Available [Online]

http://spectrolab.com/DataSheets/Panel/panels.pdf Big Brother Product Info Website: GPS Antenna. (April 2006) Available [Online]

http://www.tri-m.com/products/systems/files/specs/bigbrother_spec.pdf Advanced Micro Peripherals Ltd. Product Info Website: IEEE 1394 Interface board. (April

2006) Available [Online] http://www.ampltd.com/dload/fires2000.pdf Honeywell International Inc. Sensing and Control: Temperature Selection Guide. (April 2006)

Available [Online] http://www.ampltd.com/dload/fires2000.pdf Surrey Satellite Technologies Ltd. Spacecraft Position, Velocity and Time: GPS systems. (April

2006) Available [Online] http://www.sstl.co.uk/index.php?loc=62 Embedded Computing Design: ZF Micro Solutions. (April 2006) Available [Online]

http://www.embedded-computing.com/products/search/fm/id/?23097 The University of Hawaii CubeSat project Website: Documents (April 2006) Available [Online]

http://www-ee.eng.hawaii.edu/~cubesat/

APPENDIX A

LEONIDAS

C&DH ADS COMM POWER PAYLOAD

Height = 28 cm

Diameter = 65 cm

Antcom Antenna

QImaging UV Camera

UHM Active Antenna Big Brother

GPS Antenna

Honeywell Reaction Wheels

Xsens IMU

Parvus GPS Module

Sun Space Gyroscope

Battery Module

Tri-M Power Supply Module

Solar Panels

Aero Astro Transmitter

Yaesu UHF Transceiver

Aero Astro Receiver

AMPLTD Interface Board

Diamond Systems Interface Board

Diamond Systems Control Board

Diamond Systems Memory Board

Diamond Systems Processor/Data Acquisition

PacCom TNC

SSTL GPS Antenna

SSTL GPS Receiver