National Nanotechnology Infrastructure Network · 2019-12-19 · page i National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

page i National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

Introduction. .. .. .. .. .. .. .. .. . v

National Nanotechnology Infrastructure Network . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..vi

2006 NNIN REU Interns, by Site

Cornell NanoScale Science & Technology Facility, Cornell University .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. viiMicroelectronics Research Center, Georgia Institute of Technology .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. viiCenter for Nanoscale Systems, Harvard University .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..viiiHoward Nanoscale Science and Engineering Facility, Howard University .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. viiiPenn State Nanofabrication Facility, The Pennsylvania State University. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . ixNNIN REU Site Stanford Nanofabrication Facility, Stanford University .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . ixNanotech @ UCSB, University of California Santa Barbara .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. xMichigan Nanofabrication Facility, University of Michigan. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. xMinnesota Nanotechnology Cluster, University of Minnesota, Twin Cities .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . xiNanoscience @ UNM, University of New Mexico . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . xiMicroelectronics Research Center, The University of Texas at Austin .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. xiiCenter for Nanotechnology, University of Washington .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. xii

2006 NNIN REU Research Accomplishments Biological Applications .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..2-29

Cytotoxicity of Gold Nanoparticles in Mast Cells .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 2Eva Cornell, Physics and Economic Analysis, Gustavus Adolphus College

Microfluidic Systems for Protein Crystal Growth . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 4Kelly Costello, Mechanical Engineering, Corning Community College

Microfluidic Cell Sorters for Stem Cell Separation and Size-Profiling Using Pressurized Laminar Flows at High Spatial-Temporal Resolution. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 6

Joseph D’Silva, Electrical and Computer Engineering, Cornell University

Gold Nanoparticle-Assisted Delivery of TNF-α in Thermal Treatments of Cancer .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 8Dewi Harjanto, General Engineering / Bioengineering Concentration, Franklin W. Olin College of Engineering

Advanced Fabrication of Electroactive Nanowell Sensors .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 10Suraj Kabadi, Biomedical Engineering, Johns Hopkins University

Identifying Novel Peptides for Binding to Semiconductor Substrates to Create Nanobiomaterials. .. .. .. .. .. .. .. .. .. 12Athra Kaviani, Biomedical Engineering, University of Texas at Austin

Engineered Proteins for Binding and Organization of Inorganic Particles .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 14Elaine Kirschke, Biochemistry, University of California Santa Barbara

Nanostructure Lithography for High Throughput Cancer Screening .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 16Herbert Lannon, Physics, Rensselaer Polytechnic Institute

Microfluidic Systems for DNA Sequencing .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 18Christina Lu, Biochemistry, Brandeis University

Analysis of Experimental Data in Bacterial Adhesion .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 20Albert Mach, Bioengineering, University of California Berkeley

Single Cell Studies on Patterned, Sculptured Thin Films .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 22Robert Patrick Martin, Chemistry, Erskine College

Table of Contents

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page ii

Design and Fabrication of Nanoelectrodes for Single Cell Biosensor Applications . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 24Esha Mathew, Bioengineering, Cornell University

Patterning of Dendrimer-Like DNA 26Jon Swaim, Biomedical Engineering, University of Alabama at Birmingham

Fabrication of Surface Acoustic Wave Sensors for Early Cancer Detection . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 28Claude Wu, Electrical Engineering, University of California, Los Angeles

Electronics .. .. .. .. .. .. 30-47

Metal Nanocrystal Nonvolatile TFT Memory Cells . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 30Ravneet Bajwa, Electrical Engineering and Computer Science, University of California, Berkeley

Manufacture of Nanoscale Imprinting Stamps using Electron Beam Lithography .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 32Andrew Ballinger, Texas Academy of Mathematics and Science, University of North Texas

Fabrication and Comparison of ZnO Thin Film Transistors with Various Gate Insulators .. .. .. .. .. .. .. .. .. .. .. .. .. 34George Cramer, Electrical Engineering, The Cooper Union for the Advancement of Science and Art

Design and Testing of Microfabricated Electrostatic Fluid Accelerator .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 36Michael J. Fox, Electrical and Computer Engr, The Cooper Union for the Advancement of Science and Art

Crystallization of Amorphous Silicon Nanowires using Electromigration and Self-Heating for TFT Applications .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 38

Nathan Henry, Biomedical Engineering and Electrical Engineering, Michigan Technological University

Processing for Enabling Ultra-Fast Modulators on Chip .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 40Albert Kamanzi, Physics, University of Massachusetts at Boston

AlGaAs/GaAs Heterojunction Prosthetic Retina .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 42Juliet Lawrence, Health and Humanity, University of Southern California

Fabricating Superconducting Quantum Interference Device Nanostructures for Single Spin Detection .. .. .. .. .. .. 44Myranda Martin, Biology-Chemistry, Lock Haven University of Pennsylvania

Sidewall Metallization of High Aspect Ratio Perpendicular Polymer Structures for Chip I/O Interconnections .. .. 46Tajudeen Shodeinde, Electrical Engineering, North Carolina Agricultural and Technical State University

Materials .. .. .. .. .. .. .. 48-93

Synthesis of Semiconductor Nanoparticles .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 48Alina Ainyette, Neuroscience, Smith College

Nanoscale Materials Morphology using a Focused Ion Beam .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 50Matthew Blosser, Physics, Carleton College

Self-Assembly of Lithographically-Designed Colloidal Particles on Templated Surfaces . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 52McIntosh Bonthera, Chemical Engineering, New Jersey Institute of Technology

Nanoparticles in the Environment: A Study of Surface Reactivity of Pyrite and Arsenopyrite .. .. .. .. .. .. .. .. .. .. .. .. 54Anthony S. Breitbach, Chemistry, Clarke College

Diffusion of Aqueous Solutions in Oxycarbosilane Nanoporous Thin Films during Processing of Interconnect Structures. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 56

Sarah Bryan, Physics, Florida International University

Formation of Magnetite Nanoparticles by Thermal Decomposition of Iron Bearing Carbonates: Implications for the Evidence of Fossil Life on Mars. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 58

Alicia Cohn, Chemistry, Gettysburg College

Site Specific Nanowire Growth. .. .. 60Sonia Y. Cortes-Jimenez, Mechanical Engineering, University of Puerto Rico - Mayaguez Campus

page iii National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

β-SiC Growth on AlN-on-SiC .. .. .. 62Henry Daise III, Computer Science, Morehouse College

Rational Assembly of Semiconductor Nanowires via Dielectrophoresis .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 64Ying Yi Dang, Applied Physics, Columbia University

Measuring the Size Dependence of the Magnetic Properties of Alkanethiol-Coated Gold Nanocrystals . .. .. .. .. .. .. 66Sarah C. Hernandez, Physics and Astronomy, Texas Christian University

The Stability and Catalytic Reactivity of Colloidal Palladium Nanoparticles on Al2O3 Supports .. .. .. .. .. .. .. .. .. .. .. 68David Lavenson, Chemical Engineering, Lehigh University

Effects of N Incorporation on the Electronic Properties of GaAsN-Based Modulation-Doped Heterostructures .. .. 70Niall M. Mangan, Physics and Mathematics, Clarkson University

The Effect of Annealing Metallic Nanoparticles on their Catalytic Efficiency .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 72Katrina M. Murphy, Bioengineering, Oregon State University

Solution-Growth of Zinc Oxide Nanowires for Dye-Sensitized Solar Cells .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 74Yasuhide Nakamura, Chemistry and Chemical Engineering, University of California Santa Barbara

Cadmium Selenide Semiconducting Nanorods Vertically Aligned to a Conductive Substrate for Solar Cell Application .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 76

Emily Norvell, Materials Engineering, California Polytechnic State University, San Luis Obispo

Chemical Crosslinking and Temperature Dependant Conductivity of Ligand-Stabilized Gold Nanoparticles .. .. .. .. 78Stephanie Petrina, Materials Science and Engineering, Virginia Polytechnic Institute and State University

Epitaxial Crystal Growth of Colloids with Short Range Attraction. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 80Erica Pratt, Mechanical Engineering, Biomedical Engineering, Carnegie Mellon University

Fabrication and Characterization of Nanoporous Gold Thin Films . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 82Leila Joy Roberson, Biochemistry, Samford University

Ferrofluidic Alignment of Carbon Nanotubes. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 84Cary Smith, Physics, Jackson State University

Fabrication and Characterization of MnAs/GaAs Heterostructures for Studies of One-Dimensional Spin Transport .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 86

David Toyli, Physics, University of Minnesota

Heat Transfer Through Nanogaps. 88Josef Velten, Materials Science and Engineering, Illinois Institute of Technology

Synthesis and Galvanic Replacement Reaction of Silver Nanocubes in Organic Medium .. .. .. .. .. .. .. .. .. .. .. .. .. .. 90Kaylie Young, Chemistry, Brown University

Finding Dielectric Constant of Nanomaterials Using Fourier Transform Infrared Spectroscopy .. .. .. .. .. .. .. .. .. .. 92Caleb Yu, Electrical Engineering, California State Polytechnic University, Pomona

Mechanical Devices . 94-99

Characterization of Etching Techniques on SiC for High Temperature MEMS Applications .. .. .. .. .. .. .. .. .. .. .. .. 94Lawrence Bazille, Mechanical Engineering, Southern University

Droplet Transport Using Surface Ratchets .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 96Dane Taylor, Electrical Engineering and Physics, University of Wyoming, Laramie

Microcantilever-Based Sensors .. .. 98Amit Vasudev, Biomechanical Engineering, Stanford University

Optics. .. .. .. .. .. .. .. .. .. 100-113

Design of Mid-Infrared Ridge-Waveguide Directional Couplers by OptiWave Simulation .. .. .. .. .. .. .. .. .. .. .. .. .. 100Rose Deng, Physics, Hamilton College

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page iv

Photoluminescence of Silicon Nanocrystals Fabricated by Sputter Deposition and Annealing for Photonic Applications .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 102

Jenna Hagemeier, Physics, Northwest Nazarene University

Suspended Carbon Nanotubes for Opto-Electronic Devices .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 104Anthony Sanders, Electrical Engineering, Prairie View Agricultural and Mechanical University

Silver Nanorings: Nanofabrication and Optical Properties .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 106Julie Stern, Physics and Chemistry, Stony Brook University

Nanocrystalline Nd:YVO4 Lasing Media .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 108Chris Stoafer, Chemistry, California Polytechnic State University, San Luis Obispo

Quantum Dot Modulators .. .. .. .. .. 110Brendan Turner, Physics, Brigham Young University

Loss, Reflection and Transmission Measurement and Analysis of Silicon-on-Insulator Ring Resonators . .. .. .. .. .112Jason Wang, Electrical Engineering and Bioengineering, University of Pennsylvania

Process & Characterization .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 114-127

Research and Development of Electron-Beam Lithography Using a Transmission Electron Microscope at 200 kV .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .114

Michael Adams, B.S Physics, University of North Carolina at Chapel Hill

Fabrication of Metallic Nanoparticle Arrays .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .116Christine Chin, Mechanical Engineering, Massachusetts Institute of Technology

A Combined Electron-Beam and Nano-Imprint Lithography Technique for the Affordable Creation of Exchange Coupled Composite Patterned Media .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .118

David W. Coats, Physics, Harvey Mudd College

Fabrication of Sub-100 nm Structures Using Conventional Photolithography .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 120Paul De Andrade, Biomedical Engineering, Georgia Institute of Technology

Nanoscale Focused Ion Beam Patterning and Characterization of Perpendicular Magnetic Recording Media . .. 122Irene Hu, Electrical Engineering, Princeton University

Desorption-Ionization Mass Spectrometry on Nanoporous Polymer . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 124John Kroger, Biomedical Engineering, Georgia Institute of Technology

Fabrication of Nanometer-Scale Gaps on Thin Nitride Membranes using Electron Beam Lithography. .. .. .. .. .. 126David M. Schluneker, Mechanical/Electrical Engineering, Rose-Hulman Institute of Technology

A Fond Farewell .. .. .. .. .. 128

Index .. .. .. .. .. .. .. .. .. (Cover SEM by Michael Fox, page 36.)

129-132

page v National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

Completing a comprehensive experimental research task in 10 weeks can be a very formative experience. NNIN attempts to this by the strong collaborations and challenging tasks brought together. This report demonstrates that enthusiastic participating students coupled to the sustained support from staff, faculty, and graduate students leads to significant accomplishments.

The students participating in this effort have just started on the path of technical education and are getting their first experience with advanced hands-on research as part of our REU program. The focus on advanced research and knowledge, the strong mentoring and support, the strong exposure to a professional research environment, the strong expectations built into the research and presentations at convocations, the exposure to a wider variety of research conducted by peers and other users in diverse disciplines of science and engineering within the unifying facilities, and the strong scientific and social

interactions across the network, have been critical to the program’s success.

Equally critical is the continuing dedication and effort from our staff, faculty, and graduate students. This year’s participants also saw increased cross-site interactions through video-conferences and presentations, and hands-on experimentation. This year this final convocation happened at Cornell University.

I wish the participants the best wishes for future technical careers; NNIN hopes to see them build on this summer’s experience, and my thanks to the staff, the graduate student mentors, and the faculty for their participation and involvement. Particular thanks are due to Melanie-Claire Mallison and Lynn Rathbun at Cornell, Michael Deal at Stanford, and Nancy Healy at Georgia Institute of Technology for making their contributions in organizing and the logistics of the program and the convocation.

The 2006 NNIN REU interns at the network-wide convocation held at Cornell University, August 9-12, 2006Dede Hatch, Photographer

Introduction to the 2006 NNIN REU Research Accomplishments

Sandip Tiwari, Director, NNIN

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page vi

The National Nanotechnology Infrastructure Network is made up of the following thirteen sites, and is supported by

The National Science Foundation, the NNIN sites, our corporate sponsors and research users.

Cornell NanoScale Science & Technology FacilityCornell University

250 Duffield Hall • Ithaca, NY 14853607-255-2329 • http://www.cnf.cornell.edu

Microelectronics Research CenterGeorgia Institute of Technology

791 Atlantic Dr NW • Atlanta, GA 30332404-894-5266 • http://www.mirc.gatech.edu/

Center for Nanoscale SystemsHarvard University

17 Oxford St., Cruft 314 • Cambridge, MA 02138617-384-7411

http://www.cns.fas.harvard.edu

Howard Nanoscale Science & Engineering FacilityHoward University

2300 6th St NW • Washington, DC 20059202-806-6618

http://www.msrce.howard.edu/

Triangle National Lithography CenterNorth Carolina State University

(affiliate)

218A EGRC - Engineering Grad Res Ctr, Box 7911Raleigh, NC 27695-7920

919-515-5153 • http://www.tnlc.ncsu.edu/

Center for Nanotechnology Education & UtilizationThe Pennsylvania State University

101 Innovation Blvd, Ste 114 • University Park, PA 16802814-865-5285

http://www.cneu.psu.edu

Stanford Nanofabrication FacilityStanford University420 Via Palou Mall

Stanford, CA 94305-4085650-723-9508 • http://snf.stanford.edu/

Nanotech at UCSBUniversity of California Santa Barbara

Electrical & Computer Engineering, 5153 Engineering ISanta Barbara, CA 93106-9560

805-893-3244 • http://www.nanotech.ucsb.edu/

Michigan Nanofabrication FacilityThe University of Michigan, Ann Arbor

1301 Beal Ave • Ann Arbor, MI 48109-2122734-763-6719 • http://www.mnf.umich.edu

Nano Fabrication CenterUniversity of Minnesota

200 Union St. SE, Rm 4-174 • Minneapolis, MN 55455612-625-6608 • http://www.nfc.umn.edu/

NanoScience @ UNMUniversity of New Mexico

1313 Goddard SE, MSC04 2710Albuquerque, NM 87106

505-272-7800 • http://www.chtm.unm.edu/

Microelectronics Research CenterThe University of Texas at Austin

J.J. Pickle Research Center, 10100 Burnet Rd, Bldg. 160, Mailcode R9900 • Austin, TX 78758

512-471-6730http://www.mrc.utexas.edu/

Center for NanotechnologyUniversity of Washington

Box 351721Seattle, WA 98195-1721

206-616-9760http://www.nano.washington.edu/index.asp

REU Program Corporate Sponsors:

• Advanced Micro Devices •• Agilent Technologies •

• Analog Devices •• Applied Materials •

• Canon •• Ebara Corporation •

• Hewlett-Packard Company •• Hitachi, Ltd •

• IBM Corporation •• Infineon •

• Intel Corporation •• Intel Foundation •

• LG Electronics, Inc. •• National Semiconductor Corporation •

• NEC Corporation •• Panasonic •

• Philips •• Renesas Technology Corporation •

• Robert Bosch Corp. •• Seiko Epson Corporation •

• STMicroelectronics •• Taiwan Semiconductor Manufacturing Company •

• Texas Instruments, Incorporated •• Toshiba •www.nnin.org

page vii National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

Cornell NanoScale Science & Technology Facility, Cornell University

2006 NNIN REU Interns

Microelectronics Research Center, Georgia Institute of Technology


From left to right:

Ms. Melanie-Claire Mallison . NNIN REU Program Assistant

Mr. Herbert Lannon . . . . . . . . . . . . . . . . . . . . . . . . page 16Mr. Suraj Kabadi . . . . . . . . . . . . . . . . . . . . . . . . . . page 10Mr. Albert Kamanzi . . . . . . . . . . . . . . . . . . . . . . . . page 40Mr. Jon Swaim . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 26Mr. McIntosh Bonthera . . . . . . . . . . . . . . . . . . . . . page 52Mr. Cary Smith . . . . . . . . . . . . . . . . . . . . . . . . . . . page 84Ms. Erica Pratt . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 80Mr. Ravneet Bajwa . . . . . . . . . . . . . . . . . . . . . . . . page 30Ms. Leila Joy Roberson . . . . . . . . . . . . . . . . . . . . . page 82Mr. Nathan Henry . . . . . . . . . . . . . . . . . . . . . . . . . page 38

From left to right:

Mr. David Schluneker . . . . . . . . . . . . . . . . . . . . . page 126Mr. Claude Wu . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 28Ms. Katrina Murphy . . . . . . . . . . . . . . . . . . . . . . . page 72Mr. Tajudeen Shodeinde . . . . . . . . . . . . . . . . . . . . page 46Mr. Andrew Ballinger . . . . . . . . . . . . . . . . . . . . . . page 32Mrs. Jennifer Root . . . . . . . . . . . . . . . . . . Site Coordinator

All group photos are by Dede Hatch, Photographer

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page viii

Center for Nanoscale Systems, Harvard University


From left to right:

Mr. Paul De Andrade . . . . . . . . . . . . . . . . . . . . . . page 120Mr. Michael Adams . . . . . . . . . . . . . . . . . . . . . . . page 114Ms. Kathryn Hollar . . . . . . . . . . . . . . . . . Site CoordinatorMs. Julie Stern . . . . . . . . . . . . . . . . . . . . . . . . . . . page 106Mr. Matthew Blosser . . . . . . . . . . . . . . . . . . . . . . . page 50

Howard Nanoscale Science and Engineering Facility, Howard University


From left to right:

Ms. Alina Ainyette . . . . . . . . . . . . . . . . . . . . . . . . . page 48Ms. Juliet Lawrence . . . . . . . . . . . . . . . . . . . . . . . . page 42Mr. Henry Daise, III . . . . . . . . . . . . . . . . . . . . . . . . page 62Mr. Lawrence Bazille . . . . . . . . . . . . . . . . . . . . . . . page 94

page ix National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

Penn State Nanofabrication Facility, The Pennsylvania State University


From left to right:

Ms. Esha Mathew . . . . . . . . . . . . . . . . . . . . . . . . . page 24Ms. Sonia Cortés-Jiménéz . . . . . . . . . . . . . . . . . . . page 60Mr. R. Patrick Martin . . . . . . . . . . . . . . . . . . . . . . . page 22Ms. Myranda Martin . . . . . . . . . . . . . . . . . . . . . . . page 44Mr. John Kroger . . . . . . . . . . . . . . . . . . . . . . . . . . page 124

Stanford Nanofabrication Facility, Stanford University


From left to right:

Mr. Anthony Sanders . . . . . . . . . . . . . . . . . . . . . . page 104Ms. Maureen Baran . . . . . . . . . . . . . . . .Site AdministratorMs. Christina Lu . . . . . . . . . . . . . . . . . . . . . . . . . . page 18Ms. Irene Hu . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 122Dr. Michael Deal . . . . . . . . . . . . . . . . . . . Site CoordinatorMs. Sarah Bryan . . . . . . . . . . . . . . . . . . . . . . . . . . page 56Ms. Jenna Hagemeier . . . . . . . . . . . . . . . . . . . . . . page 102

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page x

Nanotech @ UCSB, University of California Santa Barbara


Michigan Nanofabrication Facility, University of Michigan


From left to right:

Ms. Stephanie Petrina . . . . . . . . . . . . . . . . . . . . . . page 78Mr. Jason Wang . . . . . . . . . . . . . . . . . . . . . . . . . . page 112Ms. Emily Norvell . . . . . . . . . . . . . . . . . . . . . . . . . page 76Mr. Christopher Stoafer . . . . . . . . . . . . . . . . . . . . page 108Ms. Athra Kaviani . . . . . . . . . . . . . . . . . . . . . . . . . page 12Mr. David Toyli . . . . . . . . . . . . . . . . . . . . . . . . . . . page 86

From left to right:

Dr. Sandrine Martin . . . . . . . . . . . . . . . . . Site CoordinatorMs. Ying Yi Dang . . . . . . . . . . . . . . . . . . . . . . . . . page 64Mr. Anthony Breitbach . . . . . . . . . . . . . . . . . . . . . page 54Ms. Niall Mangan . . . . . . . . . . . . . . . . . . . . . . . . . page 70Mr. George Cramer . . . . . . . . . . . . . . . . . . . . . . . . page 34Ms. Christine Chin . . . . . . . . . . . . . . . . . . . . . . . . page 116

page xi National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments

Minnesota Nanotechnology Cluster, University of Minnesota, Twin Cities


Nanoscience @ UNM, University of New Mexico


From left to right:

Ms. Dewi Harjanto . . . . . . . . . . . . . . . . . . . . . . . . . . page 8Mr. David Coats . . . . . . . . . . . . . . . . . . . . . . . . . . page 118Ms. Eva Cornell . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 2Mr. Yasuhide Nakamura . . . . . . . . . . . . . . . . . . . . page 74Ms. Kelly Costello . . . . . . . . . . . . . . . . . . . . . . . . . . page 4

From left to right:

Mr. David Lavenson . . . . . . . . . . . . . . . . . . . . . . . page 68Ms. Luxue Rose Deng . . . . . . . . . . . . . . . . . . . . . page 100Ms. Alicia Cohn . . . . . . . . . . . . . . . . . . . . . . . . . . . page 58Mr. Brendan Turner . . . . . . . . . . . . . . . . . . . . . . . page 110

Not pictured:

Mr. Josef Velten . . . . . . . . . . . . . . . . . . . . . . . . . . . page 88

National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments page xii

Center for Nanotechnology, University of Washington


From left to right:

Mr. Dane Taylor . . . . . . . . . . . . . . . . . . . . . . . . . . . page 96Ms. Elaine Kirschke . . . . . . . . . . . . . . . . . . . . . . . . page 14Mr. Albert Mach . . . . . . . . . . . . . . . . . . . . . . . . . . . page 20Ms. Kaylie Young . . . . . . . . . . . . . . . . . . . . . . . . . page 90Mr. Michael Fox . . . . . . . . . . . . . . . . . . . . . . . . . . page 36

Microelectronics Research Center, The University of Texas at Austin


From left to right:

Mr. Caleb Yu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . page 92Mr. Amit Vasudev . . . . . . . . . . . . . . . . . . . . . . . . . page 98Ms. Sarah Hernandez . . . . . . . . . . . . . . . . . . . . . . . page 66Mr. Joseph D’Silva . . . . . . . . . . . . . . . . . . . . . . . . . page 6

National Nanotechnology Infrastructure Network

Research Experience for Undergraduates Program

Research

Accomplishments

2006

BIOLOGICAL APPLICATIONS • NNIN REU 2006 Research Accomplishments page 2

BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Cytotoxicity of Gold Nanoparticles in Mast Cells

Eva CornellPhysics and Economic Analysis, Gustavus Adolphus College

NNIN REU Site: Minnesota Nanotechnology Cluster, University of Minnesota, Twin CitiesNNIN REU Principal Investigator: Dr. Christy Haynes, Department of Chemistry, University of Minnesota, Twin CitiesNNIN REU Mentors: Bryce Marquis, Department of Chemistry; Greg Haugstad, Characterization Facility;

Alice Ressler, Characterization Facility, University of Minnesota, Twin CitiesContact: [email protected], [email protected], [email protected]

Abstract:Mast cells, shown in Figure 1, are a type of immune cell found in connective tissue of the body. When the cells are stimulated they release chemicals such as serotonin from their cellular granules into the extracellular matrix.

Mast cells were cultured with various concentrations of gold nanoparticles. Amperometric techniques indicate that the nanoparticles do have a negative effect on the mast cells’ release of chemicals. Transmission electron microscopy (TEM) was used to determine the location of the nanoparticles in the cells. The nanoparticles appear to cluster in the granules near the cellular membrane of the cells.

Introduction:

Gold nanoparticles are currently being used in a wide variety of applications, from use in photodynamic therapy to use as a contrast agent. One of the unique properties of gold nanoparticles is that many of their characteristics—size, shape, and surface charge, for example—can easily be varied. Previous nanoparticle cytotoxicity studies have tended to deal primarily with live/dead assays, which

offer only a limited means of analysis of nanoparticle’s effects. A type of electrochemical analysis method called amperometry was used to study the effects of the nanoparticles on cells. Amperometry investigates the release of chemicals (such as the aforementioned serotonin) from the cell. Amperometry can be used to improve on live/dead assays and gather more detailed information about the effect of nanoparticles on cellular function. Mast cells were chosen for study since they are a model immune system cell and some of these released chemicals, such as serotonin, are electroactive and can be investigated using amperometry.

Experimental Procedure:

First, cellular function was studied via amperometry. Cells were harvested from the abdominal cavities of mice and then cultured for about 24 hours with nanoparticles of diameter 26.5 ± 6.0 nm. These cells were then stimulated with a calcium ionphore, causing degranulation of the cell and release of chemicals. The released electroactive chemicals were oxidized, causing the release of electrons to the electrode where the resultant current was analyzed to find that the addition of nanoparticles does affect cellular function. Figure 2 shows the current versus time graph for the chemical release from a mast cell that had been treated with nanoparticles. The current spikes have a much lower amplitude than that of the control group, and other key data such as half-width are also statistically different. To determine more specifically

Figure 1: TEM image of a mast cell from a mouse.

Figure 2: Amperometry data for mast cell with gold nanoparticles.

page 3 NNIN REU 2006 Research Accomplishments • BIOLOGICAL APPLICATIONS

BIO

LOG

ICA

L A

PPLICATIO

NS

how the nanoparticles were affecting the cells, the cells were then prepared for viewing using a TEM.

TEM images show cross-sectional pictures of the cells where image intensity varies according to how a beam of electrons travels through the sample. After being treated with nanoparticles, the cell pellets were fixed using glutaraldehyde and then washed in a buffer. A secondary fixation was completed with osmium tetroxide. The cells were then dehydrated using %50, %70, %95, and finally %100 ethanol concentrations. Next, the cells were placed in a transitional solvent (propylene oxide) before being moved to resin. The cells were infiltrated by resin and cured for at least three days to allow the resin to harden. A microtome was used to section 60-70 nm slices which were then stained with uranyl acetate and a modified Reynolds lead stain and viewed under a TEM at 80 kV accelerating voltage.

Future Work:

Figure 4 shows nanoparticles that have clustered in a granule of the cell close to the cellular membrane of the cell. Other images have also shown that the nanoparticles have a tendency to be located either on or directly underneath the cellular membrane of the cells. Future work could entail using atomic force microscopy (AFM) to further characterize nanoparticles located near the surface of the cell. Also, nanoparticle characteristics such as size, shape, charge, and mast cell-nanoparticle exposure time could be varied, so as to take advantage of the versatility of gold nanoparticles.

Acknowledgements:

I would like to thank Dr. Christy Haynes, Bryce Marquis, and the rest of the Christy Haynes research group. I would also like to thank Alice Ressler, Bob Hafner, Dr. Greg Haugstad, and the University of Minnesota Characterization Facility. My work was supported by the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, the National Science Foundation, the Minnesota Nanotechnology Cluster, the University of Minnesota’s Characterization Facility, and the University of Minnesota’s Electrical and Computer Engineering Research Experience for Undergraduates program.

Results and Conclusions:

Getting good quality TEM images proved to be difficult. There were problems with the hardening of the resin, the premature death of cells, a low cell yield between the cell harvest and the primary fixation of cells, and the difficult differentiation of white blood cells, mature mast cells, and immature mast cells. Once these problems were addressed, characterizing the location of the nanoparticles in the cells was fairly easy. Figures 3 and 4 show TEM images of nanoparticles in the cell.

Nanoparticles tend to cluster in the granules of the mast cells. Since the granules are essential to the primary function of mast cells, this location of nanoparticles could explain the observed negative electrochemical effect of the addition of nanoparticles.

Figure 3: TEM image of gold nanoparticles in a granule of a mast cell.

Figure 4: TEM image of gold nanoparticles in a granule near the cellular membrane.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Introduction:

The three dimensional structure of a protein determines its interactions with other proteins as well as its properties. Being able to model and understand this structure is essential in developing drug treatments, understanding genetic mutations, and creating engineered proteins with improved or desired properties. The most common method of determining this structure is using x-ray diffraction.

X-ray diffraction involves using a large, ordered protein crystal as a diffraction grating. When x-rays are diffracted, the resulting diffraction pattern can be analyzed to yield important structural information. The current rate-limiting step in this process is the crystallization of proteins. Without crystals of sufficient size and quality, the diffraction pattern is insufficient to determine the three dimensional structure.

Crystallization of proteins relies on manipulation of the phase diagram. See Figure 1. This phase diagram relates protein concentration with any one of many other variable parameters including: temperature, solute concentration, and pH. The regions of the phase diagram include the unsaturated region where proteins exist in solution, and several regions above the saturation curve where the solution is said to be supersaturated. Within the supersaturated area, there is the metastable region, then the labile region or nucleation zone, and finally

Microfluidic Systems for Protein Crystal Growth

Kelly CostelloMechanical Engineering, Corning Community College

NNIN REU Site: Minnesota Nanotechnology Cluster, University of Minnesota, Twin CitiesNNIN REU Principal Investigator: Victor Barocas, Biomedical Engineering, University of Minnesota, Twin CitiesNNIN REU Mentor: Masano Sugiyama, Chemical Engineering, University of Minnesota, Twin CitiesContact: [email protected], [email protected]

the precipitation region. In the precipitation region, an amorphous precipitate occurs, which is of no use in x-ray diffraction. The two areas of greatest concern are the labile and metastable regions. Ideally, the path taken in the crystallization process would first move into the labile region where formation of crystals occurs, and remain in the region until a desired amount of nuclei are formed. Then the solution would move to the metastable region where already formed crystals will continue to grow, but nucleation will not occur. Several methods of crystallization rely on factors such as diffusion or evaporation to create the path. The method currently used in our lab allows outside control of the movement through the phase diagram using a continuous-feed crystallization chamber. See Figure 2.

Figure 1 The rate of movement and the path taken during crystallization greatly affect the success or failure of crystal formation, but the key to crystallization of more difficult proteins lies in the nucleation stage [1]. There are two types of nucleation: homogeneous nucleation occurs within the solution, while heterogeneous nucleation occurs at the interface between the solution and its surroundings. In some cases, nucleation occurs too rapidly and produces many small crystals with too many defects for diffraction. In other cases, no nucleation occurs at all. Techniques to induce or prevent nucleation can be the key to forming large, quality crystals suitable

Figure 2


BIO

LOG

ICA

L A

PPLICATIO

NS

for structure determination using diffraction. Some of these techniques include seeding, epitaxy, or charged surfaces. The focus of our current research is inducing and controlling placement of crystal nuclei in an environment where crystals can then be easily harvested for diffraction utilizing the electrostatic charges between a positively charged protein molecule and a doped silicon surface.

Fabrication:

Continuous-feed crystallization chambers were first designed and fabricated using molded PDMS and tubing to create a new system that allows easy access to crystals formed in the chambers. Each consists of two inputs, a 2 mm x 2 mm chamber, and an output. See Figure 2. The controlled input of solution allows movement around the phase diagram relating protein concentration and solute concentration. Temperature and pH remain constant. Previous experiments were conducted in chambers made using Plexiglass® and light-cure adhesive, but the hydrophobic properties of PDMS discourage heterogeneous nucleation and allow for greater control of crystal placement as well as being less permanent.

The second stage of fabrication involves nanofabrication techniques on a silicon wafer. Starting with a p-type silicon wafer, a layer of n-type silicon is deposited on the wafer to a depth of ~ 2.5 µm. The wafer is then patterned using photolithography. The photoresist acts as a protective layer during etching, where the deposited layer of n-type silicon is selectively removed. When the resist is then removed, the result is a p-type substrate with n-type features of varying shapes, sizes, and spacings. Stripes and boxes range in size from 10-150 µm with spacings varying from 50-750 µm, while the actual growth stages are 2 mm x 2 mm for placement in crystallization chambers.


Crystallization experiments were conducted using a 35 mg/mL lysozyme solution and a 90 mg/mL salt solution with a 0.1M sodium acetate buffer at a pH of 4.2. Solutions were input at 0.5 µL/hr with outputs controlled to prevent evaporation in the chamber. Polydimethylsiloxane (PDMS) successfully prevented heterogeneous nucleation on the surface of the chamber, with crystals forming almost exclusively on the silicon growth stages. In previous experiments it has been determined that lysozyme molecules are attracted to n-type silicon at a pH level less than 7.0 and p-type at

levels above 7.0 [2]. Our preliminary results agree with this research. See Figure 3. Crystals formed selectively on the n-type features in most cases, as seen in the figure.

Future Research:

Future work is expected to include: further study of the effect of the shapes, sizes, and spacings of features; experiments conducted at varying pH levels; examination of effect of surface topography differences between deposited polysilicon versus monocrystalline silicon wafers; and tests of stages made of n-type substrate with p-type features. The hope is that the results of this research could lead to crystallization and diffraction of more difficult proteins leading to advances in pharmacology and biotechnology.

Acknowledgements:

I thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, the NSF, the Nanofabrication Center, and the University of Minnesota-Twin Cities for the opportunity; Victor Barocas and Masano Sugiyama for support and direction; Doug Ernie and Beth Stadler for a wonderful experience; Amanda Larson-Mekler, Amanda Fose, Megan Brandeland, and Paul Hattan for invaluable assistance and great company; and Larry Josbeno for introducing me to the program.

References:[1] N. Chayen. Current Opinion in Structural Biology. 14 (2004),

p.577.[2] A. Sanjoh and T.Tsukihara. Journal of Crystal Growth. 196

(1999), p. 691.

Figure 3


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:The sorting and isolation of stem cells from a medium is of great importance to the biomedical community. The focus of this project is to develop an acoustically driven, polymeric microfluidic cell sorter that will separate particles according to size-based differential migration with high spatial-temporal resolution. We modeled the acoustic field within a 150 µm high SU-8 chamber with rectangular cross sections (12.5 mm x 25 mm) sealed with 200 µm, 500 µm, and 1000 µm PDMS matching and reflective layers. We measured the energy coupled into a fluid-filled chamber of the same dimensions with 200 µm, 500 µm, and 1000 µm PDMS matching and reflective layers by the PZT to verify the model. The results show that a 200 µm thick PDMS matching layer increases the energy coupled into the chamber, while a PDMS reflective layer decreases the energy coupled into the fluid.

Introduction:

Stem cells are the future of regenerative medicine because they can develop into many types of cells. However, obtaining a stem cell “line” is often controversial because it involves the destruction of an embryo. Stem cells derived from liposuction aspirates circumvent this controversy. However, utilization of this source requires that stem cells be separated from endothelial cells and adipocyte precursors.

Microfluidic channels are promising for sorting applications: flows in microfluidic channels have a low Reynolds number, allowing laminar flow. Laminar flow prevents the shifting of suspended particles between fluid layers, thus preventing mixing of cells during sorting.

Acoustic standing waves provide an accurate, low-power method of separation that causes minimal damage to cells. These standing waves exert a primary acoustic radiation force on suspended particles (Figure 1a). In the radiation force equation (Figure 1b), if the sign of Fr is negative, the particles will migrate toward the pressure nodes, and, if positive, toward the pressure antinodes. Because of the volume dependence of the force, larger

Microfluidic Cell Sorters for Stem Cell Separation and Size-Profiling Using Pressurized Laminar Flows at High Spatial-Temporal Resolution

Joseph D’SilvaElectrical and Computer Engineering, Cornell University

NNIN REU Site: Microelectronics Research Center, The University of Texas at AustinNNIN REU Principal Investigator: Professor John X.J. Zhang, Biomedical Engineering, The University of Texas at AustinNNIN REU Mentor: Hongdae Moon, Mechanical Engineering, The University of Texas at AustinContact: [email protected], [email protected]

particles experience a greater force and reach pressure nodes and antinodes faster than smaller particles.

Simulation:

The cell sorting efficiency depends on the acoustic energy coupled into the microfluidic channel.The acoustic performance within a fluid layer of 150 µm thickness with rectangular cross-sections (12.5 mm x 25 mm) was modeled for the instances in which the energy generated by the PZT transducer was coupled through 200 µm, 500 µm, and 1000 µm thick matching

Figure 1: (a) Acoustic standing waves exert a primary acoustic radiation force on suspended particles. (b) Equations for (1) the primary acoustic radiation force and (2) the π factor.

Figure 2: (a - b) Layer configurations for acoustic energy coupling studies. (c) Fabrication of cell sorters through rapid prototyping.


BIO

LOG

ICA

L A

PPLICATIO

NS

layers of PDMS (Figure 2a). The energy density model showed a 1000 µm increase in the thickness of PDMS resulted in an energy loss of 22% and a maximum pressure and velocity decrease of 5%.

The acoustic performance within the same fluid layer was modeled for instances in which 200 µm, 500 µm, and 1000 µm thick layers of PDMS were used as reflective layers to generate acoustic standing waves (Figure 2b). The energy density model predicted that a 1000 µm increase in PDMS thickness would result in a 17% energy gain and a pressure amplitude increase of 25%.


A 150 µm thick SU-8 structure with rectangular cross-sections (12.5 mm x 25 mm) was fabricated on a 170 µm thick glass slide (Figure 2c). The SU-8 chamber was then filled to capacity with water and closed with a glass slide to which 200 µm, 500 µm, and 1000 µm layers of PDMS were attached. One PZT was then attached to the back of each of the glass slides.

In order to determine the effect of PDMS as a matching layer, the PZT attached to the glass slide covered with PDMS served as the emitter, and the PZT attached to the other glass slide served as the receiver. For each thickness of the PDMS matching layer, the energy coupled through the device by the emitting PZT was measured by the voltage generated by the receiving PZT. In order to determine the effect of PDMS as a reflector layer, the roles of the two PZTs were reversed, and the energy coupled through the stack for each thickness of PDMS was measured.

Experimental Results:

A PDMS matching layer of up to 200 µm thickness increased the energy coupled into the chamber. However, the energy coupled into the chamber decreased with increasing PDMS layer thickness beyond 200 µm (Figure 3). These results are largely consistent with the results

of the simulation, thus demonstrating that a PDMS matching layer generally decreases energy coupled into the chamber, unless the layer provides optimal impedance matching.

The energy coupled into the chamber decreased with increasing thickness of the PDMS reflector, and maximum energy was coupled into the chamber with only the glass backing acting as a reflector (Figure 4). These results show that the acoustic reflection properties of PDMS are inferior to those of glass.

Figure 3: Matching layer experimental results.

Figure 4: Reflective layer experimental results.

Conclusions and Future Work:

A PDMS matching layer of up to 200 µm thickness increases the energy coupled into a chamber by providing optimal impedance matching among the layers between the PZT and the fluid. However, matching layers of thickness greater than 200 µm decrease the energy in the fluid layer because the absorption of energy is greater than the energy gain acquired by optimal impedance matching. A PDMS reflective layer decreases energy coupled into the fluid layer by absorbing instead of reflecting the energy contained in the standing wave.

Based on these results, the design of the cell sorter can be optimized, involving the addition of a 200 mm PDMS layer above the bottom glass slide and a glass reflector plate above the channel, to achieve separation of stem cells from background particles of different sizes at high spatial-temporal resolution.

Acknowledgements:

I would like to especially thank Professor John Zhang, Hongdae Moon, and the rest of the Zhang Group. I would also like to thank the NSF, the National Nanotechnology Infrastructure Network REU, and UT MER.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract: Gold nanoparticles are seen as a promising drug carrier for anti-cancer treatments because of their size, non-toxicity, high binding capacity, inertness and stability in the human body. The effects of the gold nanoparticle-based delivery of the inflammatory cytokine, TNF-α, to breast cancer and prostate cancer cells in combination with freezing and laser-induced heating were examined. Specifically, cell viability, clonogenicity, and temperature changes were studied.

Introduction:

Gold nanoparticles are considered potential anti-cancer drug carriers for a number of reasons. Gold is non-toxic, inert, stable, and has a high binding capacity [1]; furthermore, the nanoparticles, at 33 nm in diameter, are suitably sized for laser heating [4] as well as passive uptake by the leaky vasculature of tumor tissue [1]. Gold consequently should be able to deliver drugs selectively to cancer cells. In this research, we investigated gold's potential to deliver the inflammatory cytokine tumor necrosis factor-alpha (TNF-α) to augment thermal injury selectively in tumor cells. Both freezing and laser heating were used to induce thermal injury.

TNF-α triggers an inflammatory response, causing injury both to the cells directly and the blood vessels providing them nourishment [3]. By administrating the drug using gold as a carrier, the cancer cells should become more sensitive to the thermal treatments so that less extreme parameters will be necessary, minimizing damage to the healthy tissue surrounding the tumor. To determine the effectiveness of gold nanoparticle-assisted delivery of TNF-α in conjunction with thermal treatments, two different cell types were examined. Prostate cancer cells (LNCaPs) in suspension were used for studying viability after freezing and laser heating. Breast cancer cells (SCKs) as monolayers were used for studying the effect of freezing on clonogenicity.

Gold Nanoparticle-Assisted Delivery of TNF-α in Thermal Treatments of Cancer

Dewi HarjantoGeneral Engineering / Bioengineering Concentration, Franklin W. Olin College of Engineering

NNIN REU Site: Minnesota Nanotechnology Cluster, University of Minnesota, Twin CitiesNNIN REU Principal Investigator: Professor John Bischof, Mechanical Engineering, University of Minnesota, Twin CitiesNNIN REU Mentor: Rachana Visaria, Mechanical Engineering, University of Minnesota, Twin CitiesContact: [email protected], [email protected]


LNCaPs Viability Assay with Freezing

LNCaPs were cultured in media as previously reported [2]. 1 µg/ml of native gold or gold-TNF-α was added to fresh media. Cells were incubated at 37°C with the treated media for 4 hours. Cells were detached and suspended in media. The LNCaPs were then frozen at an end temperature (ET) of -10 or –20°C using the directional solidification stage (DSS) with no hold time (HT). Viability was measured by incubating the cells with a Hoechst-propidium iodide (PI) dye for 15 minutes, after which the number of live and dead cells was counted under a fluorescent light microscope.

SCK Clonogenic Assay with Freezing

SCK cells were plated and incubated for 24 hours before being cultured with treated media (1 µg/ml of gold, gold-TNF-α, or TNF-α) for 4 hours. The treated media was then replaced with fresh media and the cells were frozen using the ethanol bath at ET –10°C with zero HT. The cells were allowed to thaw for 15 minutes at room temperature. The flasks were then left in the 37°C incubator for a week, after which the cells were stained with crystal violet and colonies were counted.

LNCaPs Temperature and Viability Assay with Laser Heating

LNCaPs were incubated at 37°C with 3.33 µg/ml Au-TNF-α-treated media for 24 hours. The cells were trypsinized from the flasks, counted, and then diluted in media until a 100 µl sample contained ~1 million cells. The samples were then irradiated for varying durations (45-75s) with a laser (532 nm, Nd:YAG) at 20 Hz corresponding to an energy density of ~ 0.4 J/cm2. Temperature was measured before and after laser treatment using a thermocouple. Viability was also assayed before and after treatment using the Hoechst-PI dye.


BIO

LOG

ICA

L A

PPLICATIO

NS


The viability results for the freeze-treated LNCaPs are detailed in Figure 1. Lower ET decreased viability. The results for the control LNCaPs and the gold-treated LNCaPs are not significantly different, confirming that native gold is non-toxic. In contrast, the LNCaPs treated with gold-TNF-α at -20°C had lower viability than the LNCaPs in the other drug treatment groups.

The SCK clonogenics data is depicted in Figure 2. Freezing significantly lowered SCK survival. The results for the control and gold-treated cells were similar, again indicating that the gold itself is not cytotoxic. Cells treated with gold-TNF-α had the lowest viability, suggesting that the gold is improving TNF-α uptake by the SCKs.

Figures 3 and 4 illustrate the results for the laser heating work. The gold-TNF-α-treated cells overall were more responsive to the heating treatment, experiencing a greater change in temperature and having lower viability than the control cells under the same laser parameters.

Lengthening the duration of the lasing treatment generally resulted in lower viabilities and greater temperature changes, with an exception occurring for the viability of the gold-treated cells with 45s laser treatment. More experiments should be performed since the data reported is only for N = 1.

Future Work:

Further work on optimizing the freezing parameters and drug dosage by evaluating cell viability in suspension can be performed. More runs for the laser heating experiments should be performed. The effectiveness of laser treatment on different cell types in suspension with varying parameters could also be analyzed before work is moved on to systems in vivo.

Acknowledgments:

The author would like to thank John Bischof, Rachana Visaria, Jing Jiang, Venkat Kalambur, and David Swanlund for their guidance and support. Special thanks are also extended to National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF, as well as to Cytoimmune Science Inc. for providing the TNF-coated gold nanoparticles.

References:[1] Paciotti G et al. "Colloidal gold: a novel nanoparticle vector for

tumor directed drug delivery." Drug Delivery. 11.3(2004):169-183.

[2] Pettaway C et al. "Selection of highly metastatic variants of different human prostatic carcinomas using orthotopic implantation in nude mice." Clinical Cancer Research. 2.9(1996):1627-1636.

[3] Visaria R et al. "Enhancement of tumor thermal therapy using gold nanoparticle-assisted tumor necrosis factor-α delivery." Molecular Cancer Therapeutics. 5.4(2006):1014-1020.

[4] Zharov V et al. "Synergistic Enhancement of Selective Nanophotothermolysis with Gold Nanoclusters: Potential for Cancer Therapy." Lasers in Surgery and Medicine. 37(2005):219-226.

Figure 3, above left: Temperature change of laser-heated LNCaPs (N=1). Figure 4, above right: Viability of laser-heated LNCaPs (N=1).

Figure 1: Viability of freeze-treated LNCaPs (N=2-6).

Figure 2: Freeze-treated SCK clonogenics (N=7).


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract/Introduction:Particle trapping using electrical fields presents as an effective way to trap particles for many biological applications. An investigation was conducted into the ability to effectively fabricate a flexible electronic architecture for use in electroactive nanowell particle trapping. This architecture consists of perpendicular fluidic channels and gold electrodes patterned into PDMS, which allows for controlled particle flow and specific targeting of wells. The lack of adhesion between PDMS and gold requires the use of a self-assembling monolayer, mercaptosilane, to effectively transfer the gold pattern. The patterned PDMS chip is then bonded to a polyimide surface, which contains micron-scale wells. By flowing polystyrene beads through this device and applying a voltage across the wells and electrodes, localized electrophoretic and electroosmotic effects can be used to trap these particles in the wells [1]. The techniques presented here can be furthered to allow for individual x-y addressability of wells and individual particle manipulation.

Fabrication Process:

The device structure consists of two independently fabricated substrates which are then bonded together. The bottom substrate is made from glass-ITO, with wells patterned in polyimide, serving as a dielectric insulator. The well sizes range from 5 to 25 µm. Fabrication of this substrate had been completed in prior research.

The focus of this project was to develop a process for fabrication of the top substrate, a PDMS chip consisting of perpendicular fluidic channels and gold electrodes (Figure 1). This was done in 3 steps: a) patterning fluidic channels on a silicon wafer, b) patterning perpendicular gold electrodes on top of these fluidic channels, and

Advanced Fabrication of Electroactive Nanowell Sensors

Suraj KabadiBiomedical Engineering, Johns Hopkins University

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: David Erickson, Mechanical and Aerospace Engineering, Cornell UniversityNNIN REU Mentor: Bernardo Cordovez, Mechanical and Aerospace Engineering, Cornell UniversityContact: [email protected], [email protected]

c) transferring the pattern in its entirety into PDMS. Standard photolithography using SU-8 10 was performed in creating fluidic channels that were 40 µm wide and 10 µm high with 200 µm spacing. In order to pattern gold electrodes on top of these fluidic channels, lift-off was performed using a very thick resist (Shipley 1075 spun at 2000 rpm) such that the layer of resist would overtop the already patterned fluidic channels. The lift-off method used was resist surface modification using toluene [2], resulting in perpendicular 40 µm gold electrodes with 200 µm spacing (Figure 2). Gold continuity was maintained along the fluidic channels by using a 2-stage evaporation at 45°, resulting in coating of the fluidic channel sidewalls (Figure 3).

In order to transfer this pattern to PDMS, the wafer was first soaked in a 15 mM solution of 3-mercapto-propyltrimethoxysilane in IPA for 2 hours, assembling a monolayer for effective PDMS-Au adhesion [3]. Sylgard PDMS (10:1) was then poured onto the wafer and cured at 80°C. After fabrication of the PDMS chip, the chip was subsequently bonded to the polyimide surface by first plasma cleaning both surfaces for 30 seconds, and then bonding the substrates so as to achieve alignment between the wells, fluidic channels, and gold electrodes.Figure 1: Diagram of top PDMS substrate.

Figure 2: Gold electrodes and fluidic channels on Si wafer.


BIO

LOG

ICA

L A

PPLICATIO

NS

Experimental Results and Conclusions:

Experiments were performed using 2 µm polystyrene beads which had been chemically treated to provide charge and fluorescence. These particles were mixed with de-ionized water and flown through the PDMS fluidic channels using a syringe pump. The first experiment conducted involved observing the effects on these particles of an applied electric field over an ITO bonding pad, i.e. a large area of ITO (1 cm2). It was observed that at relatively high voltages (around 5 to 6 V), the trapping phenomenon could consistently be observed, as all the particles over the bonding pad would come to a complete stop. This was likely due to a combination of electrophoresis, dielectrophoresis, and electroosmosis, causing the electric field lines to extend well beyond the boundaries of the gold electrodes and “trap” particles over a very large area. More importantly, this result led to the conclusion that the fabrication of the device itself was sound and gold continuity was maintained such that an electric potential can be propagated along the electrodes in the PDMS chip.

The second experiment conducted involved trying to trap these particles in the micron-scale wells, which was the primary goal of the device. Inconsistent results were observed, as the particles occasionally migrated toward the wells, but did not get trapped in them (Figure 4). Velocity changes for the particles could also be observed when voltages were applied, but not in a predictable manner. It was believed that the main reason for the unpredictable effects of the electric fields around the wells is gold-ITO contact at the ITO bonding pads, causing many of the gold electrodes to short and creating undesirable effects downstream. As a result, the device design needs to be slightly modified to prevent this from occurring.

Future Work:

In accordance with experimental results, future work will focus on slightly modifying the substrate architectures to eliminate the possibility of gold-ITO contact and shorting of electrodes. In addition, optimization of the fabrication process, particularly in regards to the concentration of mercaptosilane used and the PDMS delamination method, can contribute to improved repeatability and consistency. Ultimately, this device can be used for x-y addressability, i.e. individual well targeting, and enhanced particle manipulation.

Acknowledgements:

I would like to acknowledge the National Science Foundation and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for funding this project, and David Erickson and Bernardo Cordovez for their assistance during the course of my research.

References:[1] Cordovez, B., Psaltis, D., Erickson, D. “Trapping and storage

of particles in electroactive microwells.” Submitted 2006.[2] Campbell, S. A. (2001). “The science and engineering of

microelectronic fabrication (2nd ed.).” New York: Oxford University Press.

[3] Lee, K. J., Fosser, K. A., & Nuzzo, R. G. (2005). “Fabrication of stable metallic patterns embedded in poly(dimethylsiloxane) and model applications in non-planar electronic and lab-on-a-chip device patterning.” Advanced Functional Materials, 15(4), 557-566.

Figure 4: Unpredictable response of particles around wells.

Figure 3: Gold continuity along fluidic channel sidewall.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:Nanobiomaterials can potentially be used in a wide range of applications including, but not limited to, the development of biochips, biolabels, drug delivery systems, and bioelectronics. With the purpose of creating nanobiomaterials, we have identified peptide sequences that bind to gold substrates using bacterial surface display.

A constrained peptide library having eleven random positions (X2CX7CX2) was presented on the surface of E. coli as an insertional fusion within the extracellular loop two of outer membrane protein OmpX. Stringent bacterial selections were performed using both gold particles and gold surfaces, causing bacteria with the ability to bind gold to remain immobilized while non-binding cells are washed away. After four rounds of selection, individual clones were isolated and co-transformed with a plasmid that encodes for a green fluorescent protein. Using the fluorescent cells with displayed peptides and micron sized gold spheres suspended in PBS, it was possible to use flow cytometry to quantify cells bound to gold particles.

Clones were isolated with the ability to bind over half of the gold particles present in solution. The clone exhibiting the strongest binding to gold surfaces displayed the peptide sequence, LVCYWSYSRMCKN. This method can be extrapolated to other material surfaces, increasing our knowledge of protein-material interactions and advancing the field of protein engineering.

Introduction:

Proteins can have specific interactions with inorganic substrates [1]. These recognition properties of proteins can be harnessed to genetically engineer proteins specific for inorganics. Various display techniques have been used to screen peptide libraries for interactions with material surfaces [2]. Cell-surface display is one such biotechnology in which target proteins are stably expressed on the cell surface using transmembrane proteins allowing for peptide libraries to be screened.

Identifying Novel Peptides for Binding to Semiconductor Substrates to Create Nanobiomaterials

Athra KavianiBiomedical Engineering, University of Texas at Austin

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigator: Professor Patrick S. Daugherty, Chemical Engineering, University of California Santa BarbaraNNIN REU Mentor: Jeffery J. Rice, Chemical Engineering, University of California Santa BarbaraContact: [email protected], [email protected]


Glass surfaces deposited with gold were used for surface selections and spherical gold powder of less than 10 µm (SigmaAldrich #32658-5) for particle selections. Gold surfaces were regenerated as follows: surfaces were submersed in piranha solution (1:3 hydrogen peroxide, sulfuric acid) to remove organic residues, rinsed with acetone, isopropyl alcohol and Milli-Q water to remove other deposits, and held under a vacuum at 200°C to remove surface oxides.

A 7-mer peptide library (X2CX7CX2) was presented on the surface of E. coli as an insertional fusion within loop two of OmpX. OmpX was chosen for its long loops, fast expression, and robust display ability [3]. Stringent bacterial selections were performed using both gold particles and gold surfaces. Bacteria libraries were cultured overnight in LB chloramphenicol (Cm) and later sub-cultured. Cell cultures were induced at 0.4 OD600 for 45 minutes using arabinose, final concentration 0.02%. Cells were concentrated 40-fold using centrifugation at 5000 rpm and incubated with gold coated surfaces or gold particle suspensions (approximately 5 x 107 particles per mL). Samples added to gold particle suspensions were depleted of large cell aggregates using centrifugation at 1000 rpm for one minute. Samples were shaken for 1 hour to allow protein-gold interactions. Samples were then centrifuged at 1000 rpm for 30 seconds to remove gold particles from solution and washed with PBS to remove non-binding cells.

After four rounds of selection, individual clones were isolated and co-transformed with a plasmid that encodes for a green fluorescent protein making it possible to use flow cytometry to detect fluorescent cells bound to gold particles. Samples were sub-cultured into LB Cm carbenicillin media and grown until mid-log phase. Cells were induced for 30 minutes with IPTG (1 mM) and then induced for 1 hour with arabinose (0.04%). Cell aggregate depletions were performed at 1000 rpm for one minute. Bacterial clones were incubated in a 40 µL PBS gold particle solution with approximately 4 x 108


BIO

LOG

ICA

L A

PPLICATIO

NS

cells for one hour to allow for cell-gold interactions. Samples were centrifuged and washed with PBS before being analyzed using cytometry.


Using bacterial cells displaying a 7-mer constrained peptide library and screening techniques using gold surfaces, we successfully identified clones with a high affinity to gold substrates. After three rounds of gold-plated surface screening, and one round gold particle screening, clones were isolated with the ability to bind over half of the gold particles present in solution. Clone A is shown to bind 10.5% of gold particles in solution as determined using cytometry (Figure 1) and displays the peptide sequence, LVCYWSYSRMCKN. The lower populations in the forward and side scatter plot are bacteria cells, and the upper populations are gold particles. The gated population Gold is shown in the histogram with the gate bound cells indicating a cell/gold particle complex.

The fourth selection (Figure 2) yielded 89.5% of gold particles bound to green fluorescent bacteria cells after an extended incubation time. As seen in Figure 3, each successive round of selection provided a visible increase of the quantity of bound cells. The unselected round indicated 4.3% binding, round 1 indicated 8.3%, round 3 indicated 30.3%, and round 4 indicated 43.5%. The round 2 population exhibited a high degree of cell aggregation (data not shown). After round 2, cell aggregates were depleted as described in the methods section.

The peptide sequences identified using bacterial display through surface and particle selections have been demonstrated to confer high affinity binding to gold. With greater selection stringency, we anticipate clones of even higher affinity.

Future Work:

Our future plans include completing more surface and particle selections. We plan also to extrapolate this method to other inorganic substrates to expand our understanding of protein-inorganic interactions.

Acknowledgments:

I would like to thank Jeffrey Rice, Patrick Daugherty and the Daugherty group for their support and guidance. I would also like to acknowledge support by the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, NSF, and the Institute for Collaborative Biotechnologies.

References:[1] Sarikaya M. et al. 2003 Nat. Mater. 2, 577-584.[2] Brown, S. et al. 1997 Nat. Biotech. 15, 269-272.[3] Rice, J. J. et al. (2006) Protein Sci. 15, 825-836.

Figure 1, top: Clone A, 10.5% gold particles bound.

Figure 2, middle: Fourth Selection, 89.5% positive.

Figure 3, bottom: Selection rounds 1, 3 and 4.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:The focus of this project is the exploitation of the self-assembly properties of proteins to arrange inorganic compounds in controlled and predictable patterns. This involves the genetic engineering of the Escherichia coli DNA binding protein, lac repressor (LacI), such that it binds to inorganic compounds without significant functional loss. LacI binds with high affinity to sequence-specific regions of double stranded DNA (dsDNA), referred to as the lac operator (lacO). The addition of an inorganic binding motif would allow for the assembly of a DNA-protein-nanoparticle structure at the lacO within a complex dsDNA scaffold. Inorganic binding can be incorporated into the protein by insertion of specific polypeptide binding sequences into a previously identified permissive site in the LacI protein.

The sequence for an inorganic silica binding motif was inserted into a permissive site in LacI, following the 317th residue. The LacI derivatives were then assayed for in vivo DNA binding function and protein expression. A β-galactosidase assay determined that the LacI-silica binding derivative maintains strong DNA binding capability, and western blot analysis showed good protein expression.

Introduction:

Lac repressor controls expression of the E. coli lactose metabolism genes. This protein possesses a high binding affinity for a double stranded DNA sequence identified as the lac operon. When bound to lacO, LacI inhibits the transcription of the lacZ, lacY, and lacA genes, which encode for the proteins β-galactosidase, permease, and transacetylase, respectively. A derivative of LacI, with the ability to dually bind to DNA at a programmed location and to bind to silica, would allow for the highly controlled placement and assembly of an important material with interesting electrical properties. This technology has the potential to be highly useful in low range nanoscale ‘bottom-up’ fabrication.

A silica binding dodecapeptide (denoted QBP3) was incorporated into a LacI derivative, LacI-317, to endow the DNA binding protein with the ability to also bind to

Engineered Proteins for Binding and Organization of Inorganic Particles

Elaine KirschkeBiochemistry, University of California Santa Barbara

NNIN REU Site: Center for Nanotechnology, University of WashingtonNNIN REU Principal Investigator: Dr. Beth Traxler, Microbiology, University of WashingtonNNIN REU Mentor: Eliora Gachelet, Microbiology, University of WashingtonContact: [email protected], [email protected]

inorganic silica. LacI-317 contains a 31-codon insertion following the 317th amino acid residue, a result of Tn/lacZ/in mutagenesis, as previously described [2]. This LacI derivative maintains DNA binding functionality and thus the sequence location following the 317th amino acid was identified as a “permissive” site, a site in the protein that tolerates additional sequences without loss of protein function [3]. Insertion of the silica binding sequence into the LacI-317 gene was made possible by a unique BamHI restriction site located in the 31-codon (93 bp) insert.

Procedures:

The Tn/lacZ/in-mediated insertion mutagenesis of Lac-317 was performed on a pTrc99A derivative carrying the lacIq gene [3]. A derivative of this plasmid that contained the mutant lacI allele for LacI-317 was used as a cloning vector. This plasmid is referred to as placI-317::i31. The silica binding peptide sequence, QBP3 (Leu-Pro-Asp-Trp-Trp-Pro-Pro-Pro-Gln-Leu-Tyr-His) was originally identified in the phage display library. PCR primers (5’TTCGCAATTCCTTTAGATCTACCTTTCTATTCTCACTCT3’ and 5’ACT TTC AACAGTTTCGGCCAGATCTCCACC3’) were designed to amplify the QBP3 coding sequence with flanking BglII sites. This PCR product was purified and digested with BglII. The cloning vector was digested with BamHI. Ligation of the insert into the placI-317::i31cloning vector was followed by a BamHI digest of the ligation reaction to

Table 1: Schematic illustration of the peptides and proteins.


BIO

LOG

ICA

L A

PPLICATIO

NS

eliminate religated placI-317::i31. This product was introduced to frozen competent DH5α E. coli (F-minus, φ80dlacZ∆M15∆(lacZYA-argF)U169 deoRrecA1 endA1 hsdR17(rK

-,mK+)phoAsupE44λ-thi-gyrA96relA1), plated onto

LB-agar plates supplemented with 100 µg/ml ampicillin, and the plates incubated overnight at 37ºC.

Individual transformants where screened for the insert by PCR using an internal primer unique to the QBP3 sequence. Transformants that resulted in a PCR product were further screened by restriction digest. The binding sequence contained two unique BseRV restriction sites not found in the cloning vector, placI-317::i31. Plasmids where purified using the QIAGEN kit and digested with BseRV. Analysis by agarose gel electrophoresis reveals digested plasmids containing the insert as a linear fragment while plasmids without the insert run as uncut plasmids. Plasmids identified as containing the insert, placI-317::i31::QBP3, were transformed into frozen competent CSH140 E. coli (F128lacI/ara ∆(gpt-lac)5). The corresponding protein, LacI-317::QBP3 (see Table 1) was assayed for DNA binding functionality in vivo in a β-galactosidase activity. Since LacI represses expression of β-galactosidase, the ability of the LacI mutant to bind to the lac operon is inversely proportional to β-galactosidase activity levels. The assay was carried out as described by Kleina and Miller [3] with minor modifications. Protein expression was determined by western blot analysis, and was carried out as previously described [3], with minor modifications. The primary antibody used was IgG mouse monoclonal antibody against LacI.


Data from the insertion screening by PCR and BseRV restriction digest identified two transformants as containing the insert (Figure 1). It was determined from the β-galactosidase assay that the mutants LacI-317::QBP3 maintained excellent DNA binding (see Table 2). The western blot reveals that the mutants show significant protein expression levels, with the LacI-317::QBP3 mutant protein running at a slightly higher molecular weight compared to LacI-317 and the wild type, as expected (Figure 2).

Future Work:

The continuation of this project includes confirmation of the proper insert by DNA sequence analysis, and assaying for inorganic silica binding capability. These developed techniques could be extended to insert other inorganic binding sequences, as well as inserting the sequences into other identified permissive sites in the LacI protein.

Acknowledgments:

The National Nanotechnology Infrastructure Network REU program and NSF. Special thanks to Beth Traxler, Eliora Gachelet, Rembrant Haft, and Jessica Smith.

References:[1] Kleina, L. G., Miller, J. H. J. Mol. Biol. 1990, 212, 295-318.[2] Manoil, C., Bailey, J. J. Mol. Biol. 1997, 267, 250-263.[3] Nelson, B., Manoil, C., Traxler, B.J.Bacteriol.1997, 3721-28.

Figure 1, top: Agarose gel electrophoresis of restriction digest with BseRV.

Table 2, middle: β-galactosidase activity assay.

Figure 2, bottom: Western blot analysis.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:Optofluidic surface plasmon resonance (OSPR) is a process that can be used to detect multiple biomarkers in human sera by detection of shifts in the plasmon coupling wavelength of ordered arrays of metallic nanostructures due localized changes in the refractive index of the surrounding environment. Previously these metallic structures have been grown using “nanosphere lithography,” however our group has been working on a new technique. Using photoresist on a silicon or glass substrate, cylinders are first grown. After development, gold is evaporated onto the substrate. The cylinders are then removed and the nanostructures are left on the substrate. Currently structures of 550 nm can be made to a long-range order, but these results show there is potential for reaching 200 nm or lower. Future work will involve these structures being integrated into OSPR devices that can interrogate potentially thousands of disease markers in parallel, minimize handling and analysis time, will not require target labeling, and are inherently inexpensive to mass produce.

Introduction:

Surface plasmons are electromagnetic waves that propagate along a metal/dielectric interface, such as metal/air or metal/water. The excitation of these plasmons by light is categorized as surface plasmon resonance (SPR) for planar surfaces and local surface plasmon resonance (LSPR) for nanometer sized metallic particles. Optofluidic surface plasmon resonance is a form of LSPR. Using transmitted light and metallic nanoparticles, the coupling into a plasmon mode on the surface of the metallic nanoparticle is observed, and at a specific resonant wavelength, a dip in the transmitted power occurs. This resonant wavelength is dependent on the refractive index of the surrounding media. The SPR phenomenon can be used to detect certain bio-markers in human sera samples. When these bio-markers successfully bind, the refractive index around the nanoparticle changes and a shift in the resonant wavelength is detected.

Nanostructure Lithography for High Throughput Cancer Screening

Herbert LannonPhysics, Rensselaer Polytechnic Institute

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: David Erickson, Sibley School of Mechanical and Aerospace Engineering, Cornell University NNIN REU Mentor: Allen Yang, Chemical Engineering, Cornell UniversityContact: [email protected], [email protected]

There is significant difficulty in producing highly ordered metallic arrays of nanoparticles cheaply and easily. Without an economical and efficient way to produce these arrays; the cost effectiveness of an OSPR device disintegrates. Previous work attempted to use “nanosphere lithography,” which used close packed polystyrene spheres on a surface. Gold was then evaporated on the surface and the spheres were eliminated. Left behind was the pattern between the spheres, which can be described as six evenly spaced triangles, centered around the gap left by the sphere. Extinction spectra of this arrangement acted as predicted, but the long range order of the arrays was not very good. Our goal was to attempt to reproduce the results from this technique using standard lithography processes.


The technique used to reproduce the spheres has been used many times prior in other projects. For initial testing, a silicon substrate was used for their cost. A 500 nm layer of OiR 620-7i was hand spun on the wafer at 4000 RPM. After a pre-bake, the wafer was exposed using a GCA AutoStep 200 and a mask with 60 different circle arrays varying in size and distance apart. The circles were

Figure 1: Resist structures on silicon.


BIO

LOG

ICA

L A

PPLICATIO

NS

arranged in the same orientation as the spheres, so from a top down view, there essentially is no difference. Once exposed, the wafer was put into a YES image reversal oven. After the oven, the wafer was flood exposed at 405 nm wavelength for 60 seconds using a HTG contact aligner. The wafer is then developed in MF321 for 60 seconds leaving photoresist cylinders on the substrate, as seen in Figure 1. 5 nm of chrome and then 50 nm of gold are then evaporated on the substrate (the chrome is used for adhesion). The cylinders are finally lifted off using 1165 remover, leaving the gold arrays desired.

Future Work:

First work will need to be done to minimize the size of the nanostructures by toying around with circle size and spacing. The resist structures will have to be in the 0.6 µm to 0.8 µm range; which is a little larger the working limit of the stepper used. Next the SPR signal given by the particles will have to be maximized. This may entail changing the order of the structures, or even testing different shaped structures. Since the OSPR setup for which the arrays will be utilized is transmission based; the substrate will need to be changed to a glass one. Once all that is completed, the arrays can then be integrated into the OSPR device for more testing.

Acknowledgements:

The author would like to acknowledge and thank Professor David Erickson and Allen Yang for their guidance and insight into the project, Garry Bordonaro and Rob Ilic from the CNF for their help with fabrication, the entire Erickson Group for their support, National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for their funding, and the rest of the CNF for allowing the opportunity to work in their clean room.

References:[1] Haes, A, Van Duyne, R, “A Nanoscale Optical Biosensor:

Sensitivity and Selectivity of an Approach Based on the Localized Surface Plasmon Resonance Spectroscopy of Triangular Silver Nanoparticles,” JACS Articles, Vol. 124, No. 35, 10596 - 10604 (2002).

Figure 3: Gold nanoparticles on silicon.

Figure 2: Gold nanoparticles on silicon.


The previous arrays achieved by nanosphere lithography had an individual particle size of about 135 nm. The smallest particles in the the arrays our method achieved were about 550 nm in size. This result can be seen in Figure 2. Despite the larger size, our method displayed good uniformity over a long range. The nanosphere lithograpy method did not show this uniformity. Other configurations of cylinders were also tested, such as a square array of circles. Again the uniformity was good over the long range, but the size of the particles produced was much too big. One such result is shown in Figure 3. Another interesting result from our method was a fencing of the deposited gold. The height of the gold was clearly larger on the outside of the particles than the middle. This can be seen in Figure 2. The reason for this fencing was not determined; however it may help enhance the SPR effect. Although the particles created were not as small as initially hoped; this standard lithography process still has potential to bring the size closer to the nanosphere lithography.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:DNA sequencing is one of the most important tools in biological studies and provides a key insight into living organisms. The recent development of pyrosequencing has proven itself to be a much simpler and faster means for sequencing than traditional methods. It is currently limited, however, by its ability to produce only short read lengths. In order to make the process more efficient, we integrated the pyrosequencing technique into a microfluidic system to increase the reaction speed and also reduce the required sample size. We did this by creating a new technique that implements continuous liquid flow through microfabricated channels in silicon. We matched the dimensions of the channels to that of the capillary tubes to minimize the dead volume loss, and in doing so, hoped to achieve longer read lengths. The ability to achieve longer read lengths from the pyrosequencing technique would simplify the genome assembly process.

Introduction:

DNA sequencing has changed the dynamics of biomedical and medical studies. As more research is being conducted on the genetic make up of organisms, there is an increasing need to reduce the time, complexity, and cost of sequencing. The recent development of pyrosequencing has proven itself to be a much simpler and faster means for sequencing than traditional methods [1]. In this process, visible light is generated that is proportional to the number of nucleotides added. When polymerase adds a correct nucleotide to its base pair, the system releases a pyrophosphate molecule which is subsequently converted to ATP by the enzyme ATP sulfurylase. The ATP powers the enzyme luciferase, an enzyme commonly found in fireflies, to oxidize luciferin and produce light. This method eliminates many of the items needed for dideoxy sequencing such as labeled primers, labeled nucleotides, and gel electrophoresis.

Traditionally pyrosequencing has been conducted by separately adding the four nucleotides to DNA in a

Microfluidic Systems for DNA Sequencing

Christina LuBiochemistry, Brandeis University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford UniversityNNIN REU Principal Investigators: Dr. Peter Griffin and Professor James Plummer, Electrical Engineering, Stanford UniversityNNIN REU Mentor: Ali Agah, Electrical Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected]

test tube. Light is generated when a correct nucleotide is added, and the extra nucleotides are degraded by the enzyme apyrase. This system however, is not 100% effective and leads to a build up of by-products and residual chemicals. In addition, the system becomes increasingly dilute as more nucleotides are added. These two factors limit the length of DNA that can be sequenced, and the short read lengths make the genome assembly process more difficult.

In order to make the process more efficient, we integrated the pyrosequencing technique into a microfluidic chip. To address the problem of short read lengths, we designed the chip for continuous liquid flow, so that we could introduce fresh chemicals and flush out residual chemicals. We matched the dimensions of the chip to that of the capillary tube to eliminate dead volume areas where there could be a residual chemical build-up. By doing this, we hoped to both reduce the required sample size for sequencing as well as achieve longer read lengths.

Procedure:

We designed the microfluidic channel to have the same volume capacity as a 1/32 inch diameter capillary tube.

Figure 1: The pyrosequencing technique [2].


BIO

LOG

ICA

L A

PPLICATIO

NS

In order to complete the chip, we had to design four mask layers which required both frontside and backside alignment, as well as optimize processes for two different substrates: silicon and SU-8. First, we etched 150 µm diameter holes 100 µm deep into a silicon wafer. We then etched a 50 µm deep channel on top of the two 150 µm diameter holes. We flipped the wafer over and etched large 900 µm diameter holes 400 µm deep to meet the small holes and complete a through-hole etch.

Afterwards, we coated a glass wafer with 50 µm of the negative resist, SU-8, and proceeded to expose a single row of 50 x 50 µm wells into the SU-8. These wells served as resting areas for the DNA coated beads. These beads are introduced into the wells by drying a solution of beads over the wells. The glass wafer is then visually aligned to the silicon wafer etched with the channels, and clamped in place with a jig.

Results and Discussion:

Since we wanted to use a continuous flow system in the chip to flush out residual chemicals, we had to eliminate dead volume areas where there could be a nucleotide build up. We did this by designing the chip so that the capillary tube could be directly inserted into the chip without a connector. The diameter of the big hole was thus the same as the outer diameter of the capillary, and the diameter of the small hole was the same as the inside diameter of the capillary. The volume capacity of the capillary was equal to that of the channel.

Another unique aspect of the microfluidic chip was the use of SU-8 to form wells for the DNA coated beads. SU-8 is a hydrophobic, epoxy based photoresist. By

forming wells down to the glass we formed hydrophilic pockets in SU-8 where the beads could sit. We tested our hypothesis by drying beads on a surface and then viewing the beads with an optical microscope. We saw a high efficiency filling of the wells, indicating an affinity of the beads for the wells.

Future Work:

The next step for the project is to test the microfluidic chip by introducing liquids through a syringe and measuring the light producing reaction.

Acknowledgements:

This work has been supported by NSF through the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program at the Stanford Nanofabrication Facility and by Stanford’s Center for Integrated Systems.

I also would like to thank Dr. Peter Griffin for his helpful advisements, and Dr. Michael Deal and the rest of the SNF staff for their help.

References:[1] Ronaghi, M. Pyrosequencing sheds light on DNA sequencing.

Genome Research. 2001 Jan;11(1):3-11. [2] Nordfors, L. Large Scale Genotyping of Single Nucleotide

Polymorphisms by Pyrosequencing and Validation Against the 5’ Nuclease (Taqman) Assay. Human Mutation. 2002. 19:395-401.

Figure 2: The complete microfluidic chip. Capillary tubes are placed directly into the holes etched in the silicon wafer.

Figure 3: High efficiency beads filling of SU-8 wells.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:The receptor-ligand interaction, as mediated by the adhesive protein FimH and mannose-covered surfaces, plays an important role in bacterial adhesion. It has been suggested that the FimH receptor forms allosteric catch bonds with ligand mannose by undergoing a force-induced conformational change. Here, we use an atomic force microscope (AFM) to measure the force response of single receptor-ligand bonds, and present a computational tool for automating the calculation and analysis of the data. A computer program scripted in Matlab calculates the rupture force of single bonds and produces a bimodal figure that is consistent with the concept of two distinct, allostery bond states.

Introduction:

Understanding the molecular basis of bacterial adhesion creates opportunities for treating infectious diseases and developing novel adhesives for nanotechnology. In Escherichia coli, this adhesion is governed by a complex interplay among the receptor-ligand interaction as mediated by the adhesive protein FimH and carbohydrate mannose. Single FimH-mannose bonds were pulled apart with an AFM to calculate the force of the bond mediating this adhesion.

Calculating the rupture force and spring constant would help determine whether FimH forms allosteric catch bonds, which are longer lived bonds with increasing tensile mechanical force [1]. Current methods require days of manual hand calculations and are prone to error due to the difficulty of addressing noise in the readings. As a result, a Matlab program was written to automate the calculation and analysis of the data because it provided an efficient and accurate way to compute.

Methods:

Experimental:

To determine the force of the bond mediating bacterial adhesion, single FimH-mannose bonds were pulled apart with an AFM in contact mode. The AFM consists of a micro-scale cantilever tip coated with mannose and

Analysis of Experimental Data in Bacterial Adhesion

Albert MachBioengineering, University of California Berkeley

NNIN REU Site: Center for Nanotechnology, University of WashingtonNNIN REU Principal Investigator: Wendy Thomas, Bioengineering, University of Washington NNIN REU Mentors: Brian Kidd and Olga Yakovenko, Bioengineering, University of WashingtonContact: [email protected], [email protected]

bovine serum albumin (BSA) as well as a sample surface covered with isolated FimH complexes at the tips.

In these experiments, a stiff cantilever is pressed on the surface for a second with 100 pN force to form a bond between the tip and sample. It is then retracted from the surface at different pulling velocities. The AFM outputs binary files comprising of the force response and position of the tip.

Computational:

The goal of the Matlab program was to compute the rupture force and spring constant of the molecule consisting of fimbriae and FimH. It first detected valid adhesion events between the man-BSA cantilever tip and the fimbriae-FimH surface. An event occurred when the force increased linearly past the baseline signifying an interaction between the tip and surface in the experiment as shown in Figure 1. The baseline was created so that the program automatically filters the noise level of about 10 pN by fitting a line through the raw data and finding the intersection of certain points.

Because we would like to calculate the rupture force and molecule spring constant of single bonds, two distinct double peaks as a result of multiple bonds and null event

Figure 1: The V-shaped peak represents FimH-mannose interaction.


BIO

LOG

ICA

L A

PPLICATIO

NS

datasets were removed from further analysis. For single events or peaks, the rupture force and

distance were measured as shown in Figure 2. The rupture force is calculated as the height of the peak, or the difference between the tensile force peak and the baseline. The distance is calculated as the difference between the position of initial bond formation and the position of bond rupture.

Since pulling the cantilever and molecule acted like springs in series, we can use Hooke’s Law to compute the total spring constant, comprising of the cantilever spring constant and molecule spring constant, and then calculating the molecule spring constant using the following equations, respectively:

The molecule spring constant provides information on whether the binding adhesion is specific or non-specific, representing strong and weak forces, respectively. Non-specific adhesions are a result of low or no interaction between the tip and sample surface.

In order to run the program, the user simply inputted an initial interval for the baseline and the cantilever spring constant and chose which dataset to analyze. The Matlab program outputs a matrix containing the rupture force, total spring constant, and molecule spring constant.

Results and Conclusion:

In these experiments, a histogram of the rupture forces produced a bimodal distribution with a low force peak of 20 to 60 pN and a high force peak of 120 to 160 pN

as shown in Figure 3. The different pulling velocities suggest how the rate at which the force was increased varied. A slower pulling velocity had a higher number of low force peaks with a lower number of high force peaks while a faster pulling velocity had a lower number of low force peaks with a higher number of high force peaks.

As a result, the bimodal distribution suggests that FimH forms an allosteric catch bond. The two force peaks is consistent with the concept of two distinct bond shapes, in which FimH undergoes a conformational change for binding at a higher affinity.

However, additional calculations and further analysis are required for producing meaningful data of the molecule spring constants. Solving for the spring constant would help distinguish specific from non-specific adhesion events, which ultimately shows whether FimH forms allosteric catch bonds.

The Matlab program was fast and accurate while removing user bias. It calculated thousands of files in a few minutes and yielded similar solutions compared to manual calculations. Lastly, it offered greater insight into the FimH-mannose interaction by providing a tool to guide future experimental work.

Acknowledgements:

Many thanks to Dr. Wendy Thomas and Brian Kidd for providing an invaluable opportunity, Dr. Olga Yakovenko for gathering AFM data, Tia Ghose and Thomas lab, as well as NTUF, National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF.

References:[1] W. Thomas, et. al., “Catch-Bond Model Derived from Allostery

Explains Force-Activated Bacterial Adhesion,” Biophysical Journal, 2006, 90, 753-764.

Figure 2: Baseline = upper horizontal line. Rupture force = solid vertical line. Distance = lower horizontal line.

Figure 3: Rupture force histogram.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:In this experiment, cell studies were conducted on a patterned, sculptured thin film (STF) surface. STFs are a bionanomaterial that features a tightly-packed field of chiral, polymeric projections. This unique structure exhibits a large surface area to volume ratio; giving it significant and valuable properties as a growth medium for cells. Use of this material in transplant and biomedical events might require cells to be grown in a specific pattern. To study that phenomenon, a silicon wafer was patterned using lithography techniques, and future research will involve the transplantation of cells onto the surface of this patterned wafer.

Introduction:

Sculptured thin films (STFs) came into existence around 1994 [1]. The films are often made with a parylene derivative (typically di-chloro-di-p-xylylene) due to the well characterized biological and chemical properties of the polymer [2,3]. The basic structure of an STF is a field of chiral, polymeric projections that resemble corkscrews at the nanoscopic level. This structure, while very simplistic, allows for unique properties as a biological growth medium. The surface of the film is not smooth, presenting a large surface to volume ratio which is ideal for cell growth. The nature of the structure is also conducive to the binding of the extracellular matrix.

These unique properties of STFs make them an ideal candidate for research related to transplants and synthetic tissues. As a precursor to that research, experimentation must be conducted to study the growth patterns of cells on a patterned STF surface. The aspects of the growth that should be studied include the spacing of the pattern, growth on and off of the patterned film, and changes from normal cell growth-rates on the patterned surface versus the flat, non-patterned surface.

Methods and Materials:1. The project was begun by acquiring silicon

wafers, polished on one side and cut into approximately 1ʺ squares.

Single Cell Studies on Patterned, Sculptured Thin Films

Robert Patrick MartinChemistry, Erskine College

NNIN REU Site: Penn State Nanofabrication Facility, The Pennsylvania State UniversityNNIN REU Principal Investigator: Dr. Melik Demirel, Earth and Engr. Science and Mechanics, The Pennsylvania State UniversityNNIN REU Mentor: Eric So, Earth and Engr. Science and Mechanics, The Pennsylvania State UniversityContact: [email protected], [email protected]

2. The wafers were cleaned by acetone, isopropyl alcohol and water to remove salts and organic residues.

3. The wafers were dried with nitrogen and spun with an adhesive at 4000 rpm.

4. The wafers were all then spun in 1813 positive photoresist and baked at 115°C on a hot plate for one minute.

5. A soda-lime mask prepared by a laser writer to show squares of 30 µm on a side spaced 30 µm apart was used to expose the wafers to intense UV radiation.

6. The patterned wafers were developed with MSCD26 and blow-dried with nitrogen.

7. The wafers were cleaned in an oxygen plasma furnace and stored for later use.

8. (Future work) As the wafers are needed, they will be placed on a 2-axial motor setup, held in place by tape.

9. The wafer will then be placed in a parylene deposition system that has been loaded with 0.70g of parylene C.

10. The deposition process will then be initiated. For the deposition to take place, the furnace must be at 690ºC, the chamber gauge at 135, the vaporizer at 175 and the vacuum at 32. The last 3 numbers are given without units since they are not necessarily accurate on the instrument, but an arbitrary standard set well-beyond the actual value.


BIO

LOG

ICA

L A

PPLICATIO

NS

11. Once the deposition has finished, the wafer will be removed, cleaned by acetone, washed with water, dried and characterized. A not-to-scale example of the final product’s pattern is shown in Figure 1, courtesy of Eric So (graduate student in Dr. Demirel’s Lab).

12. Once the wafers have been sufficiently characterized, they will be shipped off to Hershey, PA, for transplanting of HEK-293 cells. The cells will then be studied for approximately 10 days. Their cell counts will be taken using fluorescence imaging and data will be collected about the location and frequency of the cells.

Results:

Results were achieved using a scanning electron microscope (SEM). Figure 2 shows a flat film. This film was produced by an improper distance (approximately 1.5 inches from substrate to funnel). This problem was fixed by reducing the distance to 0.5 inches. In Figure 3 you can see that there are columnar features. In an endeavor to minimize variables, the distance between the substrate and the funnel was reduced, but no rotation was employed-yielding a columnar surface. In Figure 4 the appropriate distance of 0.5 inches was utilized and a rotation of 0.1 revolutions per second. In Figure 2 the divide between the light spaces is a gap between the substrate and the film that resulted from a partial peeling of the film. Figures 3 and 4 do not show a clear substrate since the images were taken of a section of film that was protruding over the edge and showed a better image of the overall film.

Conclusion:

Due to several equipment failures during the time frame of the research, the full objectives of the project

could not be attained. Progress was made in showing this researcher how to prepare an STF and how to operate a variety of equipment including the scanning electron microscope, lithography tools, deposition machines, and an oxygen plasma furnace. The researcher was also able to develop analytical cleaning techniques and expand his interests in the field of biology.

Future Directions:

The future directions for this project include finishing the existing project, as well as testing other variables in the scope of this project. The distance between squares where the cells will be transplanted must be analyzed and the data has a likelihood of being cell-dependent. This dependency of the cells and the distance are other avenues of research which might be productive for transplant and synthetic tissue technologies.

Acknowledgements:

Dr. Melik Demirel and Eric So for their help with this project. The staff of the Penn State Nanofabrication Facility. The Cornell NanoScale Facility for their hospitality at the final symposium. The National Science Foundation in conjunction with the National Nanotechnology Infrastructure Network for their funding and support.

References:[1] Lakhtakia, A., and Messier, R. Sculptured Thin Films:

Nanoengineered Morphology and Optics, SPIE Press, Bellingham, WA. (2005)

[2] Demirel, M. C., So, E., Ritty, T. M., Naidu, S., and Lakhtakia, A. (2006) Journal of Biomedical Materials Research-B in press, available online at:http://dx.doi.org/10.1002/jbm.b.30656

[3] Pursel, S., Horn, M. W., Demirel, M. C., and Lakhtakia, A. (2005) Polymer 46, 9544-9548.

STFs produced at different funnel distances. Figure 2: 1.5 inches. Figure 3: 0.5 inches. Figure 4: 0.5 inches with a rotation of 0.1 revolutions per second.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:A biosensor refers to an analytical device that converts a biological response into an electrical signal. The focus of this paper is the design and fabrication of gold nanoelectrodes for the enhanced performance of cell-based biosensor applications. Proteins will be covalently bonded to the arrays of surface-engineered nanoelectrodes, over which cells will be immobilized selectively for the purpose of biosensing. The basic electrode patterning consists of 50 x 50 nm2 gold squares spaced a distance of 1 µm apart on a SiO2 wafer substrate.

Due to the small nature of the features, the electron beam lithography tool is needed to perform the exposure. In order to determine the correct charge per area to come from the e-beam, a dose array test must be carried out in which different areas of a test sample will be exposed to different charges, ranging from 250 to 800 µC. The proper dose is determined by observing which dose produces the sharpest features. Once the proper dose is established, more samples can be made.

The samples will then be sent to the University of Washington where patterning will be verified.

Design and Fabrication of Nanoelectrodes for Single Cell Biosensor Applications

Esha MathewBioengineering, Cornell University

NNIN REU Site: Penn State Nanofabrication Facility, The Pennsylvania State UniversityNNIN REU Principal Investigator: Dr. Jian Xu, Earth and Engineering Science and Mechanics, The Pennsylvania State University NNIN REU Mentors: An Cheng and Myo Thein, Earth and Engr. Science and Mechanics, The Pennsylvania State UniversityContact: [email protected], [email protected]

Introduction:

Cell-based biosensors make use of direct measurements of physiological functions and changes in function due to presence of particular substances. This provides advantages over other biosensor types because cell-based biosensors provide detection capability for previously unknown agents. This is in contrast to biosensors that, for example, use antigen/antibody binding as the detection method. With the use of antibodies, information about the substance being analyzed must be known. For the cell-based biosensor this is not necessary, as it is the response of the cells themselves that is being monitored. Applications of these biosensors include pharmaceutical screening, environmental monitoring, and toxin detection. By controlling the selective attachment of cells to these arrays, the sensitivity of these cell-based biosensors can ultimately be increased.


The first and arguably most important step in this project entailed creating a test sample upon which to run a dose array test. As the electrode patterning consisted of small features, it was necessary to determine which charge from the e-beam tool would give the desired resolution.

The substrate consisted of a 4ʺ silicon wafer, with a 2000Å layer of oxide. As the sample itself was only a quarter of a wafer, four samples could be made from each wafer. A bilayer of resists was applied to the test sample: a copolymer (MAA-MMA) and an e-beam resist (PMMA). A thin layer of 150Å aluminum was added as a conduction layer. The sample was then sent to the e-beam tool for the dose-array test. In Figure 1, a charge map of the dose-array sample can be seen. Sets of electrodes, referred to as die, are exposed to a different charge. Each die is surrounded by two finding marks, approximately 1 mm long.

Once the test sample was exposed, it was developed and the gold deposited. After the lift-off process was Figure 1: Map of dose-array test. Numerical values in µC.


BIO

LOG

ICA

L A

PPLICATIO

NS

completed, the sample was imaged under the field emission scanning electron microscope (FESEM) from which the proper charge was determined. When the optimal charge was found, more samples were synthesized.


The images taken from the FE-SEM were used to determine the proper charge per area, or dose, to come from the e-beam tool.

Shown in Figure 2 A and B are the pictures of the die and electrodes, respectively, at 300 µC. Figure 2A shows that many of the gold electrodes did not make it through the liftoff, and it was believed that this was due to underexposure, resulting in the gold being removed during the liftoff process. The electrodes that did remain had a diameter of 67 nm.

was 79 nm. Figure 4B shows that again not all the gold electrodes made it through the liftoff, this time probably due to the sonication step.

Analysis of the dose-array results determined that the best charge at which to fabricate the electrode array is 450 µC. Various steps of the sample preparation process were refined, such as the removal of the sonication step during lift-off to eliminate the chance of losing electrodes. Another important step involved the deposition of the gold electrodes via the evaporator. It was found that the evaporation level was best kept at a low value (at most 0.5 Å/sec) otherwise the copolymer layer, which is sensitive to heat, could warp and affect the electrode formation.

Future Work:

The next step will be to send the prepared samples to the collaborating group at the University of Washington to verify the patterning of the proteins and then cells upon these electrodes.

Acknowledgements:

I would like to thank the National Science Foundation, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, and The Pennsylvania State University for their funding and facilities. I would also like to thank Dr. Jian Xu, Myo Thein, An Cheng, and the staff at the Penn State Nanofabrication Facility for all their guidance and advice.

Figure 2: A) Die at 300 µC. B) Electrode at 300 µC.

Figure 3: A) Electrode at 450 µC. B) Die at 450 µC.

Shown in Figure 3 are pictures of the charge (450 µC) that gave suitable feature sizes. Figure 3A gives the diameter of a single electrode to be 72 nm.

In Figure 4, A and B are the pictures of the electrodes and die, respectively, at 600 µC. The electrode diameter

Figure 4: A) Single electrode at 600 µC charge. B) Die at 600 µC charge.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:The controlled assembly of hierarchically structured devices from nanoscale building blocks presents a great challenge to the field of nanotechnology. Dendrimer-like DNA (DL-DNA) is a versatile, nanoscale building block with a myriad of potential applications. This study presents a fabrication technique capable of dry etching poly (dimethylsiloxane) (PDMS) stamps with which lambda-DNA (λ-DNA) and DL-DNA can be patterned using molecular combing. 5 µm features were etched into PDMS, and molecular combing was employed to pattern spatial arrangements of λ-DNA. Future studies will explore the application of this process to DL-DNA in order to create multifunctional hierarchical nanostructures of DL-DNA.

Introduction:

The assembly and positioning of nanoscale building blocks in order to fabricate hierarchically structured devices remains a challenge to experimentalists. Nature, however, elegantly uses molecular recognition to fabricate supramolecular and self-assembling complexes of proteins and nucleic acids [1]. Biological molecules such as λ-DNA and DL-DNA are excellent candidates for directing the assembly and arrangement of molecular components. Controlled assembly of DL-DNA is achieved by first synthesizing Y-shaped DNA (Y-DNA), and then ligating individual Y-DNA molecules to themselves. The sequences of the Y-DNA are designed so that the ligations preferentially occur in a tree-like fashion. Higher generations of DL-DNA are constructed using the same strategy.

The objective of this project was to assemble λ-DNA and DL-DNA into patterns by virtue of a patterned PDMS template using molecular combing. In molecular combing of λ-DNA, a PDMS stamp is placed in contact with a small solution of DNA. When the PDMS is peeled off, the liquid-air interface of the traveling meniscus exerts a force on the DNA, stretching it and arranging the DNA into linear configurations. Interestingly, molecular

Patterning of Dendrimer-Like DNA

Jon SwaimBiomedical Engineering, University of Alabama at Birmingham

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Dan Luo, Biological and Environmental Engineering, Cornell UniversityNNIN REU Mentor: Wenlong Cheng, Biological and Environmental Engineering, Cornell UniversityContact: [email protected], [email protected]

combing can produce uniform arrangements of λ-DNA if the PDMS is properly patterned [2]. The patterned PDMS acts as a template, guiding DNA combing in microscale dimensions. The patterned DNA can then be transferred to glass or mica by placing and holding the PDMS stamp against a glass slide. Hierarchical structures of DNA can be achieved by transferring patterned PDMS multiple times.


Microwell arrays with a depth and diameter of 5 µm and lattice distance of 15 µm were patterned into PDMS using photolithography and reactive ion etching (REI). PDMS was spun onto silicon wafers to a thickness of 100 µm, and then cured for 90 min at 60°C. To improve the adhesion of the PDMS, silicon wafers were initially treated with air plasma (29 W) at 250 mTorr for 2 min. After curing, SPR 220-7 photoresist was spun onto the PDMS to a thickness of 7 µm and allowed to soft-bake for 30 min at 90°C. After being exposed on an EV620 contact aligner, the substrates were allowed to experience a post-exposure bake at 90°C for 30 min and then developed. Cured PDMS substrates were treated with O2 plasma (100 W) at 170 mTorr for 60 sec to improve the adhesion of the photoresist. PDMS samples were then dry etched at 43 mTorr for 30 min using a 1:3 ratio of O2 to CF4. Etch rates were optimized by varying the pressure and ratio of gases in the REI process. Etch rates were determined by attaching Kapton tape to the PDMS before etching, and then measuring the step height using profilometry. After removing the photoresist with Shipley 1165 remover, PDMS stamps were characterized with SEM and AFM. The depth of the microwell array was measured using optical profilometry.

Molecular combing experiments were performed as previously described [2]. However, an additional force of 250 mN was exerted on some of the samples as the PDMS stamps were peeled off. Fluorescence microscope images were taken during all molecular combing experiments.


BIO

LOG

ICA

L A

PPLICATIO

NS


The fluorine-based REI process developed in this study etched PDMS with a stable and directional etch rate (Figure 1). Although dry etching of PDMS has been reported previously [3], to our knowledge this is the first time anyone has demonstrated etching of features as small as 5 µm. This technique complements the standard cast-mold procedure for generating PDMS stamps and will likely be useful in the fabrication of sub-micron PDMS features. A 1:3 mixture of O2 to CF4 at 43 mTorr was found to anisotropically dry etch PDMS at a rate of approximately 10 µm per hour. Optical profilometry results revealed well depths of 4.5 µm.

Molecular combing of λ-DNA produced uniform one-dimensional arrangements of DNA. As previously reported [2], linear strands of λ-DNA were observed to attach at the edges of the wells. However, the arrangements did not possess long-range order. It is hypothesized that the treatment of PDMS with O2 plasma and/or exposure to n-methyl-pyrrolidine within Shipley 1165 remover altered the surface roughness and perhaps even the hydrophobicity of the stamp, compromising the molecular combing process. Unexpectedly, the additional molecular combing force of 250 mN was sufficient to force the DNA into the wells. Unlike λ-DNA, DL-DNA is not a linear molecular; hence, its molecular combing technique must be modified. This study presents the first evidence that our modified molecular combing technique could be employed to pattern uniform arrangements of DL-DNA.

Future Work:

Future studies should explore alternate fabrication techniques capable of producing sub-micron features in PDMS without significantly altering its surface roughness. The application of this molecular combing process to DL-DNA should also be explored in the interest of generating multifunctional hierarchical nanostructures of DL-DNA.

Acknowledgements:

I would like to thank CNF, National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, Nanobiotechnology Center, NSF, Molbel, Wenlong Cheng, Suraj Kabadi, Bert Lannon, Steve Jones and Tom Wester.

References:[1] Niemeyer, Christof M. Current Opinion in Chemical Biology

(2000) 4:609-618.[2] Lee et al. PNAS (2005) 102:51; 18321-18325.[3] Paranjape et al. JVST (2002) 20:3 975-982.

Figure 1: Anisotropically dry-etched PDMS.

Figure 2: Uniform circles of fluorescent λ-DNA using a modified molecular combing technique.


BIO

LOG

ICA

L A

PPLI

CAT

ION

S

Abstract:Surface acoustic wave (SAW) technology can be applied to create highly sensitive biosensors due to its extreme sensitivity to surface perturbation. The velocity of an acoustic wave depends upon the mass density and stiffness of the piezoelectric substrate. The binding of antigens with antibodies, when immobilized in the path of the traveling wave, changes the mass of the biolayer. The mass loading effect perturbs the surface boundary which changes the velocity of the wave and consequently shifts the frequency of the traveling SAW. With a pair of transmitting and receiving interdigital transducers (IDT), high frequency surface acoustic waves can be generated through RF interrogation. The nanoscale IDTs have been successfully fabricated using e-beam lithography. In addition, the wireless capability of the devices has been demonstrated. In the future, bio-molecule immobilization and optimization of the sensors are necessary to develop fully functional devices.

Introduction:

Recently, the applications of surface acoustic wave (SAW) devices have been extended to include bio-sensing by exploiting the extreme sensitivity of SAW to surface perturbation. SAWs are generated in piezoelectric crystals (ex. quartz) from electric signal interrogations. By adding mass density in the path of a traveling SAW,

Fabrication of Surface Acoustic Wave Sensors for Early Cancer Detection

Claude WuElectrical Engineering, University of California, Los Angeles

NNIN REU Site: Microelectronics Research Center, Georgia Institute of TechnologyNNIN REU Principal Investigator: Dr. William D. Hunt, Electrical and Computer Engineering, Georgia Institute of TechnologyNNIN REU Mentor: Christopher Corso, Biomedical Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected]

the velocity of the wave would decrease. Consequently, the center frequency shifts to a lower value.

Since SAW devices are guided by the principles similar to the Sauerbrey equation, the sensors operating at high frequency are adequately sensitive to detect mass change caused by tiny bio-molecules [1]. As shown in Figure 1, our sensor consists of a pair of input/output IDTs. SAWs of desired frequency can be generated from the input transducer, travel on a piezoelectric substrate and be received by the output transducer. With this device, a layer of bio-molecules consisting of protein cross-linkers and antibodies can be coated on the device surface in the path of the traveling waves. When the input IDT receives a pulse of an electromagnetic wave through the RF antenna, it generates traveling SAWs. If specific target proteins (antigens) are present, they bind to the antibodies, creating the mass loading on the surface of the substrate. As a result, frequency shifts in the SAWs occur. The output IDT then converts the SAWs with shifted frequency to an electric signal for RFID analysis.

The goal of this project is to fabricate this type of sensor for application in cancer detection. By immobilizing antibodies whose target proteins are specifically produced by cancer cells, we will be able to detect and identify the various types of cancer through the mechanism described.

Figure 1: Layout of the SAW biosensor. Figure 2: SEM image of the SAW device.


BIO

LOG

ICA

L A

PPLICATIO

NS

Fabrication Process:

The metal structures of IDTs were patterned using JEOL JBX-9300FS EBL system on ZEP-520A resist. Espacer 300 was spin-coated on top of the resist for anti-charging purpose. After e-beam lithography, the wafer was deposited with chromium (for adhesion) and gold. Finally, a metal lift-off process was taken to remove the resist and to form the metal structure.

Results:

The scanning electron microscopy (SEM) image of one of the best performing devices is shown in Figure 2. The two IDTs were identical, and each consisted of 40 finger pairs. The designed finger width was 324 nm and the gap between input and output IDTs was 19.14 µm. Frequency analysis data was obtained with a network analyzer connected to a probe station. We were mainly interested in measuring attenuation of the signal through the entire process.

Figure 3 shows the S21 frequency response plot of the device fabricated on a quartz substrate. The insertion loss was approximately 33 dB. The performance of devices was severely limited by the low electromechanical coupling of quartz crystal. It was also found that the center frequency of the SAW was shifted from the expected 2.44 GHz to 2.35 GHz caused solely by the mass loading of the IDT electrodes.

The frequency response can be greatly improved using a LiNbO3 substrate which is a much stronger piezoelectric material with higher electromechanical coupling efficiency. The insertion loss was approximately 13 dB for the same device fabricated on LiNbO3 (Figure 4). Again, the mass loading effect of the electrodes was observed. The center frequency shifted from the expected 2.77 GHz to 2.55 GHz. The drawback of LiNbO3 is however, a higher cost for each wafer.

On the same plot, it also shows the frequency response of the same device with the input IDT receiving the signal through a RF antenna instead of through a wired connection. In order to obtain minimum loss, we positioned the two antennas only several inches apart. The basic waveforms of the two probing methods perfectly match very closely (Figure 4). For wireless input, the signal appeared to have higher noise level due to electromagnetic interference at the wireless link.

Future Work:

Antibody immobilization should be undertaken to verify the mass loading effect of antibody-antigen binding. Further characterization and optimization of the IDTs and antennas is needed for impedance matching between acoustic devices and antennas. Creating a unique signature for each device should also be investigated to develop fully RFID-equipped devices.

Acknowledgement:

I would like to express my appreciation to Dr. William Hunt and Dr. Peter Edmonson for providing the background of this project, Christopher Corso, Ryan Westafer and the Microelectronic Acoustic Group for their mentorship and assistance, and Devin Brown and Dr. Raghunath Murali for the training on electron beam lithography. Finally, I would like to thank the MiRC, National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for providing me this research opportunity.

Reference:[1] Lee S., “Theoretical and Experimental Characterization of Time-

Dependent Signatures of Acoustic Wave Based Biosensors,” PhD thesis, Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA, 2006.

Figure 3: Frequency response of the device on quartz.Figure 4: Frequency response of the device on LiNbO3 (both wired and wireless connection).

ELECTRONICS • NNIN REU 2006 Research Accomplishments page 30

ELEC

TRO

NIC

S

Abstract:Nonvolatile memory cells consisting of a hetero-geneous self-assembled gold nanocrystal floating gate and ultra thin gate oxide were fabricated in the thin film transistor (TFT) level. The memory cells are designed to be used in flash memory applications and to improve the conventional design in terms of higher packing density, faster program/erase (P/E) operation and lower operational voltage. To achieve a controllable thin body layer, a chemical mechanical planarization (CMP) process for polysilicon was established. It was important to achieve this so that the charge stored in the gate stack layer could effectively control the channel conductance, instead of being dominated by the traps in the grain boundaries. Subsequently, the rest of the process—ultra thin oxide growth, self assembled gold nanocrystal formation, oxide and chromium deposition—was completed. In upcoming days, the devices would be characterized by measuring the memory window, P/E voltage, P/E time, retention time and cycling endurance.

Metal Nanocrystal Nonvolatile TFT Memory Cells

Ravneet BajwaElectrical Engineering and Computer Science, University of California, Berkeley

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Dr. Edwin Kan, Electrical and Computer Engineering, Cornell UniversityNNIN REU Mentors: Jaegoo Lee and Alex Tou-Hung Hou, Electrical and Computer Engineering, Cornell UniversityContact: [email protected], [email protected]

Introduction:

Figure 1 shows the structure of the metal nanocrystal nonvolatile memory cell. By applying a high voltage across the gate, electrons are forced to quantum mechanically tunnel through the thin oxide and deposit charge on the floating gate. Depending on the amount of charge on the floating gate, the threshold voltage of the device varies. Two such states with sufficiently large threshold voltage difference, known as memory window, can therefore be used to represent the binary logical states. Currently, there are two major limitations in the conventional memory cell design: program/erase (P/E) operation takes considerably long (1-10 ms) and P/E voltage is too high (12-19V). These could however be improved if the tunneling oxide thickness is reduced but the scaling down of the existing design is limited due to high leakage current which degrades the retention characteristics. Thus, an improvement in the design is needed. This project was aimed at fabricating and characterizing an improved design with two major distinctions: using gold nanocrystals as a discrete floating gate which would reduce the leakage current as oxide thickness is reduced, and using thin polysilicon film as the transistor substrate which would enable 3-D integration of memory cells and provide compact embedded memory for devices.


The first step of the fabrication process was to obtain about 20 nm thick polysilicon film with RMS roughness below 2 nm. To achieve this we had to establish a CMP process using the Strasbaugh 6EC CMP tool.

A two-level four-variable statistically designed experiment was thus set up. The four variables (values chosen) under testing were: down force (4 and 6 psi), table speed (15 and 25 rpm), chuck speed (15 and 20 rpm), and slurry type (Semi-Sperse® P1000 and Semi-Sperse® 12). The thickness and roughness of the wafers were measured before and after the CMP process, and the

Figure 1: Structure of a memory cell.

page 31 NNIN REU 2006 Research Accomplishments • ELECTRONICS

ELEC

TRO

NIC

S

removal rate for each setting was calculated. The settings were then compared to choose the best combination. After establishing this process, fabrication of our device wafers was started.

About 500 nm of PECVD oxide was deposited on a silicon wafer which was followed by a 100 nm undoped polysilicon deposition. This polysilicon was then polished down to about 20 nm by using the established CMP process. The next step was to grow 3 nm of oxide and it was followed by about 1.2 nm of gold deposition. The gold self-assembled into nanocrystals (6 nm in diameter) owing to high surface energy difference.

Next, about 30 nm PECVD oxide was deposited and finally 200 nm of chromium was evaporated on the wafer to serve as the gate electrode.


The first interesting fact learned while establishing the CMP process was that native oxide grown on polysilicon, even a few angstroms, considerably reduced the overall removal rate when P-1000 slurry was used. Consequently, the process was adapted by adding a HF dip step to remove the native oxide immediately before polishing the wafers, thus making the CMP process more predictable. Removal rate and uniformity of different settings used in the design of experiment (DOE) are summarized in Table 1 and 2. Clearly, setting 4 had the least removal rate and most uniformity. Also, as shown

in Figure 2, AFM roughness results revealed that RMS roughness of polysilicon decreased down to an acceptable value of 1.29 nm from about 4.05 nm and therefore this setting was chosen. Using this setting, smooth polysilicon films as thin as 15 nm were successfully obtained and processed further to make devices. Finally, the wafers were sent out for ion implantation to enhance the conductivity of the drain and the source terminals.

Future Work:

The devices will soon be characterized by measuring the memory window, P/E voltage, P/E time, retention time and cycling endurance. The design will then be optimized until we meet our goals of faster and low power-consuming flash memory.

Acknowledgements:

This study was funded by NSF through National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program at Cornell NanoScale Facility, Cornell University. I would like to thank Dr. Edwin Kan, Jaegoo Lee, Alex Tou-Hung Hou, CNF staff, and the rest of the Kan group.

Figure 2: AFM image of polished polysilicon.

Table 1, above: CMP data using P-100 slurry.

Table 2, below: CMP data using SS-12 slurry.


ELEC

TRO

NIC

S

Abstract:This project has focused on characterizing all of the processes leading from a bare silicon wafer to the ultimate use of stamps to pattern interconnects for electronics on the 20-100 nm scales. We created our initial patterns in hydrogen silsesquioxane (HSQ) using e-beam lithography. We then used reactive ion etching (RIE) to transfer the patterns onto a silicon wafer. This wafer was coated in a fluorinated anti-stick layer and then used as a stamp to transfer interconnect-patterns into PMMA.

Introduction:

Electron beam lithography (EBL) is a key process that has been used to fabricate nanoscale patterns on a variety of substrates. Though e-beam lithography is a very powerful tool in the laboratory, the process is far too slow to be used in industry.

Nanoimprinting lithography is a method that has recently been investigated to bring the small scale patterning capabilities of e-beam lithography to speeds acceptable for mass-manufacture.

Miniaturization has been the key force driving the constant increase in computing power that we have enjoyed over the past sixty years. As we continue to reduce the scale of our computing devices, we are forced to develop new techniques and strategies. Photolithography reaches its theoretical limit at ~ 32 nm using current proposed methods.


The fabrication process was broken up into three main steps; each step would undergo a series of optimizations so that the result could be used to optimize the next step in the process.

We started our process by spinning a layer of HSQ, a negative e-beam resist, onto the surface of a 4 inch silicon wafer. At this stage of the process there were a number of parameters we experimented with to produce better

Manufacture of Nanoscale Imprinting Stamps using Electron Beam Lithography

Andrew BallingerTexas Academy of Mathematics and Science, University of North Texas

NNIN REU Site: Microelectronics Research Center, Georgia Institute of TechnologyNNIN REU Principal Investigator and Mentor: Devin Brown, Senior Research Engineer,

Microelectronics Research Center, Georgia Institute of TechnologyContact: [email protected], [email protected]

results. We used an e-beam pattern with line gratings of differing line thickness and pitch (Figure 1) and a scanning electron microscope (SEM) to optimize our doses for minimal backscattering. During this step we also tried different resist thicknesses and development conditions.

The next step in our process focused on plasma processing. The goals of this step were to reduce the linewidth further by using an isotropic CF4 chemistry, and then to successfully transfer the pattern from the resist into the silicon substrate using an anisotropic O2 plasma chemistry. By varying the plasma chemistry and then observing the results under SEM, we were able to adapt existing chemistries to our need. By reducing the DC bias and increasing the chamber pressure, we produced a more isotropic CF4 etch. We also determined the etch rates of each plasma chemistry and used this data to vary stamp thickness in our next step.

We also experimented with an SF6 linewidth reduction etch, but the silicon was being etched with the resist and the results were poor. We decided to focus on the CF4 etch due to time concerns.

The next phase of our procedure was the nanoimprinting step. Features on this scale could only be imprinted if they had been successfully fabricated in an earlier processing step. In this step we used an Obducat nanoimprinter. To prepare our stamp for imprinting, we spun on a thin layer of a fluorinated anti-stick. To prepare our stamping substrate, we spun an oxide wafer with a polymethyl

Figure 1: 1000 nm lines with 1:1 pitch are bottom left, single pass lines with 50 µm pitch are top right.


ELEC

TRO

NIC

S

methacrylate (PMMA) film. We used the oxide as an adhesion layer between the silicon and the PMMA. We varied the thicknesses of the various layers, as well as imprinting temperature, pressure, and time. We observed the results under SEM.


Using e-beam lithography we managed to achieve sub 10 nm lines with some consistency; below 10 nm, lab conditions prevented accurate imaging. We determined that a higher concentration developer for a longer development time provided a sharper contrast curve.

Using a CF4 plasma chemistry, we were able to reduce the line thickness from 11 nm to 7.8 nm. The CF4 linewidth reduction quantity was shown to be a linear function of time, which would allow the overall thinning to be controlled.

Using a silicon etch we were able to confirm the presence of 30 nm features in silicon for use as imprinting stamps. We found that our silicon etch had an etch rate of 29 Å/second and a selectivity of silicon to HSQ of 8.6 to 1.

The nanoimprinting step presented the most difficulty; especially because it could only be attempted after the previous steps had been refined. We were able to imprint 81 nm lines with 500 nm pitch.

Future Work:

Future work would consist of further optimization of each step. Nanoimprinting would then be used to fabricate sub 32 nm interconnect by producing an imprinted pattern with an e-beam dosage and development conditions calculated using our first step of optimization, a plasma chemistry refined in our second step, and an imprinting process used in our last step. The resulting negative would then go through a lift-off process to produce a testable interconnect pattern.

Acknowledgments:

I would like to thank my research mentor and PI, Devin Brown, for his guidance. I would also like to thank Raghunath Murali and Joel Pikarski for their training and various insights into the field of nanotechnology research. Special thanks goes to Jennifer Tatham Root and her incredibleness. Of course, thanks to the entire MIRC staff, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF.

Figure 2, above: Zeiss SEM of an 11 nm line in HSQ using the JEOL electron beam lithographer.

Figure 3, below: Zeiss SEM image of 81 nm lines with 500 nm pitch.


ELEC

TRO

NIC

S

Abstract:High electron mobility and visible transparency make ZnO thin film transistors (TFTs) an attractive alternative to amorphous silicon TFTs for use in flat panel displays. The insertion of a high-κ gate dielectric, (Pb,Zr)TiO3 (PZT), into ZnO TFT structures was compared to SiO2 and SiNx gate dielectrics with the goal of improving switching characteristics and achieving lower operating voltages. ZnO thin films were deposited by pulsed laser deposition, PZT by a sol-gel process, and SiO2 and SiNx by PECVD. ZnO TFTs utilizing PZT insulators demonstrate significantly improved gate control, with some degradation in drain current.

Introduction:

A thin film transistor is a type of field-effect transistor that provides a common, inexpensive method of driving individual pixels in LCD and OLED displays. Increased drain currents resulting from high mobility are necessary for achieving high frame rates in high-resolution displays. Low deposition temperatures are essential for deposition on inexpensive, transparent substrates, such as glass and plastic. Zinc oxide is an ideal TFT semiconductor for these applications, having a higher mobility than amorphous silicon and lower deposition temperatures than polycrystalline silicon. Also, the high bandgap of ZnO (3.3eV) leads to a desirable level of transparency of the transistor itself. The use of high-κ dielectrics in TFTs

Fabrication and Comparison of ZnO Thin Film Transistors with Various Gate Insulators

George CramerElectrical Engineering, The Cooper Union for the Advancement of Science and Art

NNIN REU Site: Michigan Nanofabrication Facility, University of MichiganNNIN REU Principal Investigator: Jamie Phillips, EECS Department, The University of MichiganNNIN REU Mentor: Jeffrey J. Siddiqui, EECS Department, The University of MichiganContact: [email protected], [email protected]

is becoming increasingly necessary, as smaller transistors lead to unacceptable levels of gate leakage current. A high-κ PZT (κ ~ 100) dielectric replacement for SiO2 (κ = 3.9) would serve to lower transistor operating voltages and allow the insulator thickness to be maintained, thus reducing gate leakage without sacrificing capacitance. This research compares the performance of ZnO TFT samples with three insulators: PZT, SiO2, and SiNx.


ZnO was deposited using pulsed laser deposition (PLD). In this process, the substrate was placed in a vacuum (10-6 Torr) and heated to 350°C. Under 30 mTorr O2, the target was struck by a 248 nm KrF excimer laser pulsed at 350 mJ, 20 ns pulse width, and 6 Hz. The ZnO target was ablated, forming a plasma plume that deposited 80 nm of ZnO on the substrate. PZT was deposited using a sol gel process. PZT nanoparticles were spun onto the substrate in a colloidal suspension. A soft bake and rapid thermal anneal were used to evaporate the carrier solvent and to enhance crystallinity in the deposited thin film. A final thickness of 240 nm was measured. TFTs were fabricated on Pt/SiO2/Si substrates, which provided a common metal gate for all transistors (Figure 1). First, the insulators were deposited, PZT by the sol gel process, SiO2 (240 nm) and SiNx (150 nm) by PECVD at 400°C. Next, ZnO was deposited using the PLD technique.

Standard photolithography, metallization, liftoff, and wet chemical etching processes were used to form source/drain contacts and to access the Pt gate. The aspect ratio fabricated for these transistors was W/L = 113 µm/ 14 µm. A subsequent BHF etch was carried out to remove ZnO between each TFT, thus electrically isolating the devices. Current-voltage DC testing data was gathered using the Keithley 4200-SCS, while capacitance-voltage data was measured using a Boonton capacitance meter at 1MHz.

Figure 1: TFT schematic and SEM.


ELEC

TRO

NIC

S


The SiO2 voltage sweep yielded results very similar to that of SiNx, shown in Figure 2. Compared to both oxide and nitride, the PZT TFT works at much lower voltages, as shown in Figure 3. In addition, the PZT sample transitions from an “on” state to an “off” state with only a small change in gate voltage. This is evident by the small subthreshold slope given by Figure 4. These characteristics are related directly to the PZT sample’s relatively high capacitance attributable to the high dielectric constant of the PZT material. The PZT sample also shows significantly lower drain current than either oxide or nitride. This undesired result coincides with the unexpectedly low value of mobility. Since the mobility is an intrinsic function of the ZnO channel between source and drain, this value should not change with different insulators. One possible cause would be a roughness of the PZT surface, which might hinder electron transport through the thin ZnO layer.

The ZnO mobility is very low even for the SiO2 and SiNx samples, given that a ZnO mobility of 27 cm2/V.s

has been reported for similar ZnO deposition conditions [1]. The low ZnO mobility is likely a result of non-optimal ZnO PLD deposition conditions. The PZT TFTs began showing significant gate leakage current for gate voltages above 4V, possibly due to physical defects in the material associated with the sol gel process. Alternately, the low PZT bandgap (3.4 eV) relative to ZnO may not provide an adequate energy barrier to effectively impede current flow.

Future Work:

Adjust the PLD operating conditions to improve the ZnO crystal structure and maximize ZnO mobility. The resulting greater “on” current will improve the on/off current ratio. Also, attempt to reduce gate leakage, especially with PZT. Gate isolation helps to achieve this, but the current etch method can damage the device. Fabricating an independent gate for each transistor would provide an alternative to this etch. Gate leakage might also be reduced by investigating the use of thin, high-bandgap dielectric layers between ZnO and PZT. In addition, determine the level of physical uniformity of the PZT layer. A smoother PZT surface might reduce gate leakage and improve the effective channel mobility. Finally, attempt to reduce the “off” current by depositing a thinner layer of ZnO. This would increase the source/drain resistance and allow for easier depletion of charge in the semiconductor.

Acknowledgments:

Thanks to Prof. Jamie Phillips, Jeffrey J. Siddiqui, Emine Cagin, Patrick Chan, Sandrine Martin, the staff at the Michigan Nanofabrication Facility, and to NSF and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Fortunato, E; Barquinha, P; Pimentel, A; “Wide-bandgap

high-mobility ZnO thin-film transistors produced at room temperature”; Applied Physics Letters, 85(13), 2541-2543 (2004).

Figure 4: Summary of results.

Figure 2, above: Silicon nitride TFT drain sweep.

Figure 3, below: PZT TFT drain sweep.


ELEC

TRO

NIC

S

Abstract:The micro cooling industry requires novel cooling solutions to compensate for heat generated by increasingly component-dense electronic and MEMS devices. Rotary fan technology is unable to produce adequate fluid flow and is not readily scaleable. Electrostatic fluid acceleration (EFA) provides an energy efficient high performance alternative that has the potential to be manufactured directly into the electronics themselves. Proof of concept micro EFA devices were fabricated and tested for their operational I-V characteristics. Geometric optimizations were incorporated in design, and their effects demonstrated experimentally.

Introduction:Rapidly shrinking electronics and MEMS devices generate

heat from electrical resistance, mechanical friction, and even combustion processes. Existing thermal management solutions have inadequate heat removal and energy efficiency characteristics to meet industry needs.

Conventionally, heat dissipates away from a hot substrate through a thermally conductive heatsink. Fluid flow then removes this heat to the atmosphere. The traditional rotary fan fails to offer the performance and flexibility demanded by advanced thermal management applications. High speed moving parts produce unwanted vibration and noise and are difficult to scale down. Resulting cross sectional air velocity profiles are static and experience a dip in the center opposite the motor assembly.

Electrostatic fluid acceleration which ionizes and accelerates an atmospheric fluid in a desired flow pattern is a promising replacement for the removal, or fluid flow, stage in micro cooling. EFA technology has been around for some time, and has been investigated for use in air filtration, propulsion, humidification, propulsion, and audio speakers. For the cooling of microelectronic and MEMS devices, EFA offers several important advantages over the rotary fan: near laminar air profile with controllable velocities, simple solid-state electrode geometry which offers excellent design flexibility and ease of manufacture, low power consumption, large convective heat transfer, and the possibility to decrease the effective boundary layer at the solid fluid interface within an enclosed structure

Design and Testing of Microfabricated Electrostatic Fluid Accelerator

Michael J. FoxElectrical and Computer Engr, The Cooper Union for the Advancement of Science and Art

NNIN REU Site: Center for Nanotechnology, University of WashingtonNNIN REU Principal Investigator: Professor Alexander Mamishev, Electrical Engineering, University of Washington NNIN REU Mentors: Nels Jewell-Larsen and Chih-Peng Hsu, Electrical Engineering, University of WashingtonContact: [email protected], [email protected]

like a channel. The mechanisms of EFA action by corona discharge based ionization have been previously investigated and modeled with respect to electrostatic, fluid dynamic, and space charge effects [1]. Practical micro EFA devices were fabricated and tested for their operational voltage and current characteristics in the first ever foray into the development of an integrated, or on-chip, EFA device.

Background:The mechanism of EFA action is illustrated in Figure

1. Voltage is applied between a high tip curvature corona electrode and a low tip curvature collector electrode. High electric field intensity in the vicinity of the corona electrode ionizes air molecules. These ions are accelerated towards the collector, transferring kinetic energy to surrounding neutral molecules. The EFA has three regions of operation. At low levels of applied voltage, the EFA is cut-off and has zero current. At some onset voltage the EFA current begins to increase exponentially with applied voltage. The EFA operates between the onset voltage and the breakdown voltage, where dielectric breakdown occurs between the corona and collector electrodes, causing sparking between the electrodes and poor net fluid flux.

A wide range of different air velocities and voltage characteristics have been reported for different devices. The variables of import are the corona surface area, air gap size, and corona electrode curvature. Experiments at the macro and meso scale have investigated the variation of these parameters and what follows is the application of understood optimizations to micro devices.

Figure 1: Ion wind from EFA operation.


ELEC

TRO

NIC

S

tests to ensure devices conformed to desired dimensions and did not change during operation or from breakdown effects.

Micro devices were tested to determine the dependency of onset and breakdown voltages on tip length as well as air gap size between electrodes. Positive voltage was applied to the Corona electrode and the collector was held at a lower voltage. Current data points were taken for a particular device geometry and air gap size throughout the cut-off and operational voltage range up until the expected point of breakdown or appearance of sparks. Figure 4 shows the effects of variations in air gap size on the operational voltage range of a 5 mm long tip. Opera-tional voltage range increased along with the air gap size.

Experimental data, not shown here, indicates that onset voltage is lowered and operational range extended for devices with longer tips. This was observed via extensive testing of 3, 5, and 8 mm tip lengths. Preliminary tests indicated air velocities for some devices in excess of 2 m/s.

Future work includes further miniaturization and characterization of micro EFA devices moving towards a fully optimized single wafer on-chip solution.

Acknowledgements:The guidance and mentorship of Professor Alexander

Mamishev, Chih-Peng Hsu and Nels Jewell Larsen was greatly appreciated as was the donation of time and resources by NTUF, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF.

References:[1] N. E. Jewell-Larsen, P. Q. Zhang, C. P. Hsu, I. A.

Krichtafovitch, and A. V. Mamishev “Coupled-physics modeling of electrostatic fluid accelerators for forced convection cooling” IAA/ASME June 2006.

Figure 2: Corona electrode geometry and SEM image.

Figure 3: Ultra high tip curvature electrode.Figure 4. Current-voltage curves for EFA operation

at various electrode separations.

Procedure:Devices were fabricated from bulk silicon using industry

standard photolithographic techniques. Lowering the necessary onset voltage for EFA operation was the principle design goal. Figure 2 shows the geometry of the devices alongside an SEM image of an actual device. The wafers were etched using deep reactive ion etching (DRIE).

High curvature was desired at the very tip of the long cantilever on the underside of the wafer. The overall tip length serves to isolate the tips’ electric field from that of the low curvature base. The negative sidewall tapering (top width > bottom width) was achieved by patterning large spacing around the cantilever for masking during the DRIE. Devices were fabricated with varying tip length, height, top width, and bottom width. Etch times were continually modified to establish reliable expectations of tip dimensions and curvature. Figure 3 shows a device with sharpening by RIE for high tip curvature.

Results:The devices were tested to obtain current-voltage

relationships for micro EFA device operation at known corona electrode to collector electrode orientations and separation distances. The devices were held at a distance above a semiconductive foam collector electrode by a micro-positioning x-z stage. SEM imaging was used before and after


ELEC

TRO

NIC

S

Abstract:Large area electronics, such as displays and touch screens, employ a network of thin film transistors (TFTs) to control image circuitry. The TFTs are commonly built on amorphous silicon (a-Si), providing small device-to-device variation, even though its mobility is much lower than other forms of silicon. Our goal was to investigate the possibility of crystallizing a-Si nanowires to form single crystalline areas by creating a single crystalline domain along a nanowire through self-heating and electro-migration techniques by application of electrical current. Employment of this technique in nanowires, rather than large area films makes it possible to achieve single crystal domains per wire rather than a polycrystalline structure, which would result in much higher carrier mobility while maintaining small device-to-device variations.

Introduction:Displays require a high degree of uniformity in pixel

arrangement to form desired images. This requires very small device-to-device variations in the device used for the driver circuitry and lighting element.

Thin film transistors, used in display technology, are fabricated from amorphous silicon (a-Si), due to uniformity of the material and overall cost, when compared with poly-Si and crystalline-Si fabrication. These two forms of silicon are not commercially used, but have greater electron mobility, as seen in Figure 1; making them superior candidates for future transistor fabrication.

Our solution to the problem of low mobility in a-Si was to use self heating and electro-migration to create a single crystal domain along a nanowire. Crystallizing a-Si reduces the interstitial boundaries between the atoms, decreasing electron collision and increasing mobility.

Crystallization of Amorphous Silicon Nanowires using Electromigration and Self-Heating for TFT Applications

Nathan HenryBiomedical Engineering and Electrical Engineering, Michigan Technological University

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator & Mentor: Ali Gokirmak, Electrical and Computer Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

When applying current to the nanowire, the internal temperature increases, allowing the embedded hydrogen to release and silicon atoms to re-arrange into a more orderly structure; beginning crystallization. Electro-migration is the bombardment of material atoms by electrons, causing detrimental erosion over time. By using this mobile force, we attempted to direct crystallization along the nanowire, resulting in a single crystalline domain. Preliminary experiments on making use of self-heating and electro-migration were investigated.

Experimental Procedure:An array of nanowires connecting 100 µm x 100 µm

contact pads were designed in a computer added design (CAD) program. The array ranged from 50 nm to 440 nm widths with 10 nm increments and lengths varying from 1.5 µm to 5.5 µm.

The nanowires were fabricated by depositing 200 nm of N-doped a-Si on a 776 nm silicon oxide, thermally grown on a silicon wafer. SPR-955 photo resist was spun at 4000 rpm for 60 seconds. Using the Auto Step (I-line optical stepper) and the designed mask, the wafer was exposed for 0.09 seconds resulting in nanowires that were 2x the size. The wafer was baked at 115°C for 90 seconds. Etching was performed in the Oxford 80; RIE using CF4 was the process used.

After fabrication, electrical current was passed through individual nanowires. An initial I-V measurement was taken to understand the preliminary electrical conductivity of the nanowire. A constant voltage was then applied to the wire, ranging in bias and time duration based on wire dimensions. This voltage application was repeated many times, to determine what effect current had on the nanowire. A final I-V measurement was taken for electrical comparison. Electrical testing was completed when the wire was either deemed broken or near the breaking threshold.

Results and Conclusion:After electrical current had been passed, a noticeable

discoloration was first observed. The nanowire and electrode pad changed from a blue hue to a pink tint; the nanowire being most vibrant. The electrode pad measuring incoming voltage remained unchanged. Discoloration of the a-Si is probably due to the occurrence of oxidation.

We examined the nanowires using a scanning electron Figure 1: Mobility comparison among different silicon material.


ELEC

TRO

NIC

S

microscope (SEM) to verify whether the nanowires were broken or intact. Cracking was observed to extend into the pad areas, shown in Figure 2, which suggested significant electro-migration or internal mechanical stress. Another observation was the formation of white bubbles along the side of the nanowire, which could not be explained. Some texture change was observed in the nanowires around the crack regions, but no crystallization was seen.

In order to increase flexibility within the nanowires, silicon oxide was etched beneath the N-doped a-Si using buffered oxide etch (BOE) solution, removing 500 nm of silicon oxide. This etching resulted in suspended beams. Electrical testing was performed again on various nanowires.

Results of the testing provided interesting findings. The first nanowire to show significant crystallization was a 250 nm BOE nanowire. An SEM image of the nanowire, seen in Figure 3, showed considerable grain alteration on the side closest to the applied voltage. The crystallization is visible by color dissociation. Crystallization did not progress along the entire length of the nanowire, but instead stopped where the nanowire broke, due to material disparities.

Complete crystallization along the whole length of the nanowire can be seen in Figure 4. The nanowire broke when 30 µV was applied over a fraction of a second. This was a mishap that occurred due to forgetting to reset the voltage from a previous test. The excessive voltage increased the temperature above 1414°C, high enough to melt the a-Si; and cooled rapidly when the current flow stopped. The nanowire solidified into a smooth rounded structure, tapering to the breakage point. The smoothen nanowire is distinguished from the surrounding non-crystalline material due to a change in surface granulation. This nanowire is the best candidate for single grain crystallization and TEM imaging of the atomic structure would provide verification.

The experiments on these nanowires are presently used to investigate the time scales and the necessary voltage which should be applied to form crystalline nanowires while keeping the wires attached the contact pads.

Acknowledgements:I would like to thank my PI and mentor Ali Gokirmak for

his strong dedication and influence in helping me to learn. Many thanks to my research site at Cornell University, and Melanie-Claire Mallison for making sure the wheels of success ran squeaky free. Thanks to National Nanotechnology Infrastructure Network for funding this great research opportunity of which I was most appreciative to be a part.

References:[1] Jon Revie and Travis Nelson; “Thin-film Transistor LCD

Display”; http://www.cs.ndsu.nodak.edu/~revie/amlcd/index.html#Concl (1996).

[2] “Learn about LCD TV and TFT LCD Displays”; http://www.plasma.com/classroom/what_is_tft_lcd.htm (2002).

3] Wikipedia; “Liquid Crystal Display”; http://en.wikipedia.org/wiki/Liquid_Crystal_Display (2006).

Figure 2, top: SEM image of 200 nm RIE etched nanowire.

Figure 3, middle: SEM image of 250 nm BOE nanowire.

Figure 4, bottom: SEM of 140 nm crystallization of nanowire.


ELEC

TRO

NIC

S

Abstract:In order for ultra-fast electro-optic modulators to perform faster [1], low resistance ohmic contacts are required. Several methods have been developed to achieve these contacts. For silicon chips, transition metals are made to react with the silicon at the contact points, forming silicides. The most used silicide is TiSi2 due to its low resistivity. For very thin layers (nanoscale) of silicon on insulator (SOI), this technology faces challenges with the difficulty to obtain TiSi2, with high resistance TiSi being formed instead. We focus on forming contacts using both titanium and nickel for silicidation. The metal deposition and annealing times are carefully controlled and varied to determine the best contacts on thin silicon layers of 50, 100 and 150 nm.

Introduction:

The electro-optic modulator under consideration is made up of waveguides and a p-i-n diode, as shown in Figure 1. The diode must be of minimum thickness so as to not affect the light propagating in the waveguides.

Processing for Enabling Ultra-Fast Modulators on Chip

Albert KamanziPhysics, University of Massachusetts at Boston

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Prof. Michal Lipson, School of Electrical and Computer Engineering, Cornell UniversityNNIN REU Mentor: Carl Poitras, School of Electrical and Computer Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Formation of metal contacts then presents a difficulty because it involves the formation of silicides on thin silicon, around 50 nm. The silicidation is an important process because it allows for a gradual change in chemical potentials from the silicon to the top metal pads, and therefore requires investigation.

In this project, TiSi2 and NiSi are separately investigated and are formed on thin SOI. Titanium is used because TiSi2 has low resistivity [2]. Also Ti is very effective at reducing native oxides in contacts. But a major drawback is that it can form another higher resistivity silicide, TiSi. By carefully monitoring the annealing process and the amount of metal used, these two silicides can be controlled. This is the main aim of the project. With nickel, only one silicide is formed (NiSi). The formation of NiSi consumes less silicon than in the TiSi2 process, and is therefore more appropriate for use with thin layers of silicon.


The devices were built on three SOI wafers with 50, 100 and 150 nm of silicon respectively. The various devices included resistors, diodes and sheet and contact resistance test structures (including structures which use the transfer length method (TLM). See Figure 2).

Figure 1: Electro-optic modulator developed by

the Lipson group [1]. Figure 2: TLMs that were used to measure resistance as a function of separation distance.


ELEC

TRO

NIC

S

Photolithography was used to define the layers onto the wafers using the GCA Autostep 200 stepper. The active regions were implanted with boron (n+) and arsenic (p+) to concentrations of about 1019 cm-3.

A 1 µm layer of silicon dioxide was then deposited onto the wafers by plasma-enhanced chemical vapor deposition. The contacts were then defined also using the Autostep, followed by a SiO2 etch using the Plasmatherm 72 fluorine etcher. The contact metal (Ti or Ni) was then evaporated onto the wafers after they underwent a quick HF dip. The annealing step followed using a rapid thermal anneal (RTA) tool. Chips with Ti were annealed at 400°C for 180s, then at 650°C for 120s and then at 800°C for 30s. As for Ni, the anneal was done at 500°C for 30s. A last photolithography step defined the metal pads (for laboratory testing), and Al was evaporated.

Results And Conclusions:

The TLMs were used for measuring the contact resistance of the silicides. This technique allows for the measurement of contact resistance between neighboring pads that are separated by a gradually increasing distance, using a 4-probe setup. The contact resistance is extrapolated for a null separation distance (see Fig. 3).

Future Works:

Ni appears to offer the lowest contact resistance for this type of technology. It does require a low temperature anneal which was performed using an RTA outside of the clean room. If Ti is still to be considered for the silicide, the use of this outside tool should be considered since it is reliable at low temperatures—as compared to the RTA tool in the clean room which is accurate only at temperatures above 600°C. Several of the required anneals are below this temperature.

A recipe which was expected to give better results for titanium contacts was not tested because of time limitation. This would involve depositing titanium at a thickness of only 20% of that of the silicon layer. Using less Ti could prove beneficial because when excess Ti is present, it reacts with the formed TiSi2 giving TiSi. Another important step would be to anneal the titanium chips at temperatures above 850°C. At this temperature, TiSi2 converts its phase from C49 to a lower resistivity phase, C54 [3].

Acknowledgements:

I would like to thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, the National Science Foundation and the Intel Foundation for making this possible. I would also like to thank the Cornell NanoScale Facility staff for their help. Special thanks to my Principal Investigator Prof. Michal Lipson and my mentor Dr. Carl Poitras.

References:[1] Xu, Q., Schmidt, B., Pradha, S. and Lipson, M., “Micrometer-

scale silicon electro-optic modulator”, Nature, Vol. 435, 325-7 (2005).

[2] Gambino, J.P. and Colgan, E.G., “Silicides and ohmic contacts”, Materials Chemistry and Physics, Vol. 52, 99-146 (1998).

[3] Iwai, H., Ohguro, T., Ohmia and A.-I. “NiSi salicide technology for scaled CMOS”, Microelectronic Engineering, Vol. 60, 157-69 (2002).

Table 1 shows the extrapolated contact resistance for chips with NiSi. The values are much lower than for the Ti contacts, where measurements were about 10-20 times higher. Ni definitely offered the lowest contact resistance. It is possible that the Ti contacts did not form a good silicide.

Figure 3: Example of measured resistance as a function of distance between measuring probes for a given TLM.

Table 1. Contact resistance for the NiSi contacts.


ELEC

TRO

NIC

S

Abstract:Diseases such as retinitis pigmentosa and age-related macular degeneration lead to gradual loss of eyesight due to the progressive loss of retinal photoreceptors. Currently, several treatments for these diseases are being used to slow vision loss. One in particular hopes to restore partial vision by implanting an artificial retina using solar cells to provide electrical stimulation of the ganglion cells of the eye when exposed to light.

The objective of this work was to fabricate an AlGaAs/GaAs prosthetic retina with an array of 10 µm diameter solar cells on a 3 x 3 mm2 chip. Open circuit voltages of 0.82V were obtained under illumination for these cells. Mesas were etched into both sides of this chip to minimize electrical crosstalk between cells and minimize movement of the chip once implanted.

Introduction:

Once light enters the eye, it stimulates the retina’s rods and cones. These are photoreceptors which are responsible for detail, low light and color vision. This light is then converted into electrical signals which are sent to the brain, via the optic nerve, and interpreted as vision. There are several retinal diseases such as retinitis pigmentosa (RP) and age related macular degeneration (AMD) which disrupt the normal functions of the eye, more specifically the retina. RP is a genetic disorder which causes abnormalities of the retina’s photoreceptors. AMD is a disease which causes loss of the sharp central vision needed to read or drive.

Since RP and AMP affect millions of people worldwide, strides are being made to slow down or stop their progression. Many researchers and practitioners believe that administering beta carotene (vitamin A), vitamins C, E and other nutrients can lower the risk of developing or slow down the progression of particular retinal diseases. There are also many FDA-approved drugs such as Macugen® and Visudyne® that are used with

AlGaAs/GaAs Heterojunction Prosthetic Retina

Juliet LawrenceHealth and Humanity, University of Southern California

NNIN REU Site: Howard Nanoscale Science and Engineering Facility (HNF), Howard UniversityNNIN REU Principal Investigator: Dr. Gary L. Harris, Director, HNF, Electrical Engineering, Howard UniversityNNIN REU Mentor: Mr. James Griffin, Senior Research Associate Lab Manager, HNF, Electrical Engineering, Howard UniversityContact: [email protected], [email protected]

photodynamic therapy (PDT) to treat patients. Lastly, many labs are developing artificial retinal implants (prosthetic retinas) to restore vision to patients. One very promising type of implant utilizes solar cell technology to provide electrical stimulation to the retina.

Many researchers, like those at Optobionics, are using silicon (Si) to make solar cell implants. Though silicon-based implants have several advantages, GaAs-based solar cells are more efficient. By adding aluminum to GaAs, the spectral response can be tuned to resemble that of the human eye (Figure 1). A thin layer of AlGaAs near the surface helps to decrease surface recombination while serving as a window allowing light to penetrate other regions of the cell.

Experiment:

The fabrication of the prosthetic retina began with the growth of GaAs and AlGaAs layers by molecular beam epitaxial (MBE) on a 100 µm thick n+ GaAs wafer. The first layer was a n+ GaAs buffer layer followed by a lower doped n+ GaAs layer. The third layer grown was p+ GaAs in order to create a p-n junction needed to power the cell. Next, a thin layer of p+ AlGaAs was grown and followed by a very thin cap layer of p+ GaAs. All n-type and p-type layers were doped with silicon and beryllium respectively.


ELEC

TRO

NIC

S

After growth, metal contacts were evaporated onto both sides of part of the wafer to test for cell efficiency. The bottom n-type contact was 20 nm of germanium : 20 nm of nickel : 200 nm of gold. The top p-type contact, which were 10 nm wide fingers, were 40 nm of titanium : 30 nm of platinum: 100 nm of gold. The contacts were then alloyed at 450ºC for 1 min. Electrical measurements indicated a short circuit current of 2.1 mA and an open circuit voltage of 0.82V. The fill factor was calculated to be 0.52 and the solar cell efficiency was 5%. Fabrication of the prosthetic retina also involved making contacts to both sides. The bottom side contact was the same as mentioned above while the top contact was a 7 nm thick layer of gold. The top contact was made semi-transparent to allow light to penetrate the cell.


Previously, solar cells made from epilayer structures similar to the one used in this work had yielded efficiencies as high as 20%. The 5% obtained in this work is well below what was expected. The reason for this low efficiency is not known. The good news is that a high open circuit voltage of 0.82V was obtained for this structure. It should be noted that the efficiency of the final prosthetic device could not be measured because the 10 µm cells were too small to contact.

In the future, RIE conditions need to be optimized to allow deeper and cleaner mesas. The AlGaAs/GaAs epilayer structure needs to be optimized to produce a higher efficiency. Redesign of the retina mask should also be done to more closely simulate the natural eye retinal geometry. Finally, an encapsulating layer will need to be developed before the device can implanted into the eye.

Acknowledgements:

The author would like to thank the National Science Foundation and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for funding, and everyone at the Howard Nanoscale Science and Engineering Facility for the research.

References:[1] “eye, human.” Encyclopedia Britannica. 2006. Encyclopedia

Britannica Premium Service. 21 July 2006.[2] “Fabrication and Characterization of AlGaAs/GaAs

Haterostructure Photovoltaic Cell for Prosthetic Retina Applications.” Ewart Orr.

[3] www.optobionics.com. [4] http://science.howstuffworks.com/solar-cell.htm.[5] “GaAlAs/GaAs Heterojunction Prosthetic Retina.” Stephanie

Cheng. 2003 NNUN Research Accomplishments.

The next step was to form mesas on the top surface by reactive ion etching (RIE). The area that was not etched (an array 10 µm diameter cells with a 50 µm pitch) was masked with either a thick photoresist or metal layer by photolithography. The device was then exposed to 15 sccms of Cl2 at a pressure of 15 mtorr and 225 watts of RF power for 30 minutes. These conditions produced a mesa depth of approximately 10 µm with vertical sidewalls (Figure 2). After RIE, any remaining photoresist or metal mask was removed and the sample alloyed at 450°C to form ohmic contacts.

The final prosthetic retina was 2.5 mm2 in size and consisted of roughly 1900 solar cells and is shown in Figure 3 next to a dime to provide a size comparison.


ELEC

TRO

NIC

S

Abstract:The overall goal of this project was to fabricate extremely sensitive microbridge superconducting quantum interference devices (SQUIDs) [1] using two high resolution electron beam resists stacked in a bi-layer. The geometry of the microbridge SQUID is needed because this particular type of SQUID will be used to detect the spin of single molecule magnets. In order for the SQUID to be more sensitive than current microbridge SQUIDs, the specific goal and main focus was to create lines 20 nm in width in the junction of the SQUID rings (see Figure 1). The SQUID rings were created using the Leica EBPG-5HR electron beam lithography tool.

In the making of the SQUID devices two types of metal were used; niobium and aluminum. Niobium was deposited through sputtering, and aluminum through thermal evaporation. Images of the SQUID rings were then taken using a field emission scanning electron microscope (FESEM). Finally, the SQUID samples were sent to the University of California, Berkeley, for characterization.

Fabricating Superconducting Quantum Interference Device Nanostructures for Single Spin Detection

Myranda MartinBiology-Chemistry, Lock Haven University of Pennsylvania

NNIN REU Site: Penn State Nanofabrication Facility, The Pennsylvania State UniversityNNIN REU Principal Investigators: Dr. Khalid Eid and Dr. Jeffrey Catchmark,

Penn State Nanofabrication Facility, The Pennsylvania State UniversityContact: [email protected], [email protected]


The project began with a 3ʺ bare silicon wafer that was cleaned with de-ionized water, acetone and isopropyl alcohol to remove any undesired residues. The wafer was then prepared by spinning and baking two high resolution electron beam resists in a bi-layer configuration. The wafer was then exposed to the electron beam lithography tool (e-beam) to make a dose array. From the dose array, the proper dose needed for the 20 nm lines in the SQUID rings would be determined. After exposure to the e-beam, the wafer was developed. Different developing schemes were used to determine the correct developing times in the three different chemicals used in the developing process. The wafer was developed in n-amyl acetate, methyl-isobutyl ketone:isopropyl alcohol (8:1), and de-ionized water. Metal was then deposited onto the wafer and a lift-off process took place. The final step was imaging the SQUID rings with FESEM. From the images obtained, the proper dose and developing scheme were determined.

Another 3ʺ bare silicon wafer was cleaned, prepared, and processed with the determined dose from the e-beam and developing times. For the metal deposition step in the procedure, the wafer was cleaved in half to create two samples of SQUID rings. Niobium was sputtered onto one half and aluminum was thermally evaporated onto the other. For each sample, the thickness of the metal deposited was 10 nm. During the lift-off procedure, ultrasonic agitation was used to promote the lifting-off of the metal. The two samples were then imaged with field emission scanning microscopy and sent to the University of California, Berkeley, for characterization.


From the dose array, an exposure of 180 µC/cm2 was found to be sufficient for the formation of the 20 nm lines in the junction of the SQUID ring. The determined

Figure 1: FESEM image of a SQUID ring.


ELEC

TRO

NIC

S

developing scheme was 90 seconds in n-amyl acetate, 30 seconds in methyl-isobutyl ketone:isopropyl alcohol (8:1) solution and 30 seconds in de-ionized water. The niobium SQUID sample did not lift-off as well as the aluminum (see Figure 2). This may possibly be due to the fact that sputtering is less collimated than thermal evaporation. Ultrasonic agitation did help promote the lift-off in both of the samples. Using the determined exposure and developing scheme from the dose array, SQUID rings having junction lines approximately 20 nm in width were successfully created in the final sample (see Figure 3).

Future Work:

The future work for this project includes testing that will be conducted at the University of California, Berkeley. At this location, the SQUID rings will be tested to determine whether they are more sensitive than current microbridge SQUID devices, and whether they are capable of detecting single spins or not. They will then be used to study spin dynamics in single molecule magnets. Other future work includes the “fine tuning” of the recipe to create the microbridge SQUID devices.

Figure 3: FESEM image of junction of SQUID with 20 nm lines.Figure 2: FESEM image of niobium SQUID with poor lift-off.

Acknowledgements:

I would like to thank the National Nanotechnology Infrastructure Network and the National Science Foundation for their funding and Penn State University for the use of their nanofabrication facility. I would also like to thank my Principal Investigators Dr. Khalid Eid and Dr. Jeffrey Catchmark for their guidance and support, as well as John McIntosh, Guy Lavallee, and Michael Rogosky without whom this project would not have been possible. In addition, I would like to thank Lisa Daub, Robert Ehrmann, and Lucas Passmore for an enjoyable and educational experience at Penn State University.

References:[1] K. Hasselbach, C. Veauvy, and D. Mailly, Physica C:

Superconductivity 332, 140 (2000)


ELEC

TRO

NIC

S

Abstract:As complementary metal oxide semiconductor integrated circuits (CMOS ICs) continue to advance, the need for ever better chip input/output (I/O) interconnect technology becomes ever more critical. In this research, we explore the use of a “Sea of Polymer Pillars” to transmit electrical and optical signals rapidly from the die to the substrate. The intrinsic polymer pillars (or optical I/Os) are processed through microelectronic fabrication techniques that include spin coating, soft baking, exposing to UV light, hard baking, developing, and curing. Mechanical reliability plays a major role due to the fact that these pillars must be compliant enough to withstand the thermo-mechanical stresses induced as a result of the difference of the coefficient of thermal expansion (CTE) of the die and the substrate. On the other hand, before these polymer pillars can actually transmit electrical signals, the pillars need to be metallized with an electrical conductor, copper, on the sides and not the tips. The primary focus of this research is to develop a unique process for single sided metallization of polymer pillars to allow them to operate as electrical and optical I/Os.

Introduction:

Chip I/O interconnects provide mechanical inter-connection between the silicon die and the substrate. They also provide a path for delivering power supply current, high frequency signals and heat dissipation [1]. There are various types of compliant I/O interconnects including micro-springs, helix-like interconnects, coil-like interconnects, and “Sea of Polymer Pillars” [2].

“Sea of Polymer Pillars” is an I/O interconnect technology that represents modern day micro-electro-optical-mechanical systems (MEOMS) [3], in the sense that they are very small structures that transmit electrical and optical signals while undergoing mechanical deformations. These polymer pillars are mechanically compliant cylindrical structures that mitigate thermo-mechanical expansion mismatches between the chip and

Sidewall Metallization of High Aspect Ratio Perpendicular Polymer Structures for Chip I/O Interconnections

Tajudeen ShodeindeElectrical Engineering, North Carolina Agricultural and Technical State University

NNIN REU Site: Microelectronics Research Center, Georgia Institute of TechnologyNNIN REU Principal Investigator: Dr. Muhannad Bakir, Electrical Engineering (Microelectronics), Georgia Institute of TechnologyNNIN REU Mentor: Calvin R. King Jr., Electrical Engineering (Microelectronics), Georgia Institute of TechnologyContact: [email protected], [email protected]

substrate. A scanning electron microscope (SEM) image of such pillars can be seen in Figure 1. Our research goals were partitioned into two tasks. The first task of this research project involved fabricating the polymer pillars and metallizing one side of their sidewall (instead of the complete sidewall) and not the tips with a good electrical conductor, such as copper (Cu), so that the electrical-optical pillars would be more flexible.

It is known that the CTE of the silicon die and a substrate have a large mismatch. While the CTE of the silicon die is 3 ppm/°C, the CTE of the printed wiring board (PWB) is 17 ppm/°C. Thus, this indicates that when these two materials undergo thermal cycling, the substrate will have a greater expansion than the silicon die, leading to bending of the I/O interconnections. Therefore, the last half of the project included making these I/O interconnections (polymer pillars) more compliant by increasing their high aspect ratio.

An illustration of the problem can be seen in Figure 2, while the goal is illustrated in Figure 3. In addition, we discovered that we can utilize our unique process technology to fabricate solely very high aspect ratio (> 20:1) vertical copper wires by thermally decomposing the polymeric material. While this result was not part of

Figure 1: SEM micrograph of 100 µm tall polymer pillars.


ELEC

TRO

NIC

S

the original research plan, it was a critical demonstration as it could potentially have very significant applications in a number of areas.


Two layers of Avatrel polymer were spun on an oxidized wafer. The first layer was pre-baked for 8 minutes at 108°C before the second layer was spun on the sample. The second layer of Avatrel was soft baked for 75 minutes at the same temperature to reduce any solvents. After the sample cooled, lithography was performed on the sample. Next, the wafer was hard baked for 20 minutes and the sample was developed to produce the required polymer pillars. The pillars were cured for one hour. Once the pillars had been cured, thin layers of titanium and gold were uniformly deposited. A second set of polymer pillars were fabricated on the metallized surface using the same recipe. During the lithography step, the mask was aligned so that the patterns on the mask covered half of the polymer pillars on the sample. The sample was then hard baked and developed. The first layer of titanium covering the gold layer was etched using buffered oxide etchant (BOE). Next, copper was metallized on the sidewall of these pillars by electroplating. This step concluded the first half of the project.

To proceed with the last half of the project, after the sidewalls of the pillars had been metallized, the sample was heated in a furnace at 450°C to thermally decompose the polymer. After the polymer had been decomposed, the gold and the titanium seed layers were etched using gold and BOE etchants, respectively, to produce the high aspect ratio copper pillars (Figure 4).


The pillars were fabricated and their sidewalls were metallized with copper. Figure 4 illustrates how the sample looked after the polymer had been decomposed when heated in the furnace.

The sidewall metallization remains as very high aspect ratio copper pillar (> 20:1).

Future Work:

Future work includes optimizing the fabrication process of the structures under consideration and testing the mechanical, electrical, and optical properties of the electrical and optical polymer pillar I/Os. Assembly of these I/Os on a PWB should also be performed in the future.

Acknowledgments:

The author would like to strongly thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for funding this program. Additionally, I wish to thank Georgia Institute of Technology, Ms. Melanie-Claire Mallison, Dr. James Meindl, Mrs. Jennifer Tatham Root, Dr. Muhannad Bakir, Microelectronics Research Center (MiRC) staff members, GSI members, GSI administrative staff members and Mr. Calvin R. King Jr, for the acceptance, knowledge, enjoyment, guidance as well as mentorship in this summer research.

References:[1] M. Bakir, et al, Electronic Components and Technology

Conference, 54, Vol 1, pp 1-6, 2004. [2] M. Bakir, et al, IEEE Photonics Technology Letter, Vol 16, No

1, pp 117 - 119, 2004.[3] R. Villalaz, “Volume Grating Couplers for Optical Interconnects:

Analysis, Design, Fabrication, and Testing”.

Figure 2: Short I/O interconnects break during thermal cycling.

Figure 3: Long I/O interconnects bend instead of break during thermal cycling.

Figure 4: SEM images of sidewall metallized polymer pillars before and after polymer decomposition.

MATERIALS • NNIN REU 2006 Research Accomplishments page 48

MAT

ERIA

LS

Synthesis of Semiconductor Nanoparticles

Alina AinyetteNeuroscience, Smith College

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard UniversityNNIN REU Principal Investigator: Dr. Gary Harris, Engineering, Howard UniversityNNIN REU Mentor: Dr. Jason Matthews, Chemistry, Howard UniversityContact: [email protected], [email protected], [email protected]

Abstract:Three secondary amines, N-methylaniline, diisopropylamine and diethylamine, were reacted with carbon disulfide in the presence of diethylzinc to yield the corresponding bis (dialkylamidodithiocarbamato) zinc complexes in yields ranging from 10 to 24%. Benzoic acid was reacted with diethylzinc to afford bis (benzoate) zinc in 62.7% yield. The metal complexes were characterized via FT-IR spectroscopy. The precursors were then combined with trioctylphosphineoxide and trioctylphosphine (TOPO/TOP) and thermally decomposed to afford ZnS and ZnO nanoparticles respectively. The nano-particles were evaluated via UV-Vis spectroscopy and light scattering. The λmax for ZnO was 279 nm, which was different from that of the bulk material (370 nm). The λmax range for ZnS was 272-287 nm, which was different from that of the bulk material (340 nm).

Introduction:

Nanoparticles have recently attracted significant attention from the materials science community. Nanoparticles, particles of material with diameters in the range of 1 to 20 nm, promise to play a significant role in developing technologies [1]. They exhibit unique physical properties that give rise to many potential applications in areas such as nonlinear optics, luminescence, electronics, catalysis, solar energy conversion, and optoelectronics. Two fundamental factors, both related to the size of the individual nanocrystal, are responsible for these unique properties. The first is the large surface to volume ratio, and the second factor is the quantum confinement effect [2].

The synthesis of single source precursors for use in the preparation of semiconducting nanoparticles is of significant interest to the materials scientist in that it allows for excellent control of product stoichiometry. Nanoparticulate zinc sulfide has recently been targeted for use in electronics. Synthesis of this zinc based material has been previously achieved via a three step process where the salt of a dialkylamidodithiocarbamate is

prepared and subsequently made to undergo a metathesis reaction with a zinc halide to give the single source ZnS precursor, zinc bis(dialkylamidodithiocarbamate). The third and final step is the decomposition of the prepared zinc bis(dialkylamidodithiocarbamate) in the presence of trioctylphosphine (TOP) and trioctylphosphine oxide (TOPO). Herein, we describe our results from a two step route where the zinc bis(dialkylamidodithiocarbamate) is prepared from the reaction of a secondary amine with carbon disulfide (CS2) and diethylzinc (DEZ) (Figure 1). The isolated zinc bis(dialkylamidodithiocarbamate) is then thermally decomposed in TOP/TOPO affording the ZnS nanoparticle (Figure 2).

General Procedures:

Synthesis of bis(dialkylamidodithiocarbamato) zinc: Synthesis of the zinc complexes were carried out by combining 5 mmol of the appropriate secondary amine with an equimolar amount of carbon disulfide in pentane (20 ml) in a round bottom flask. Next, 2.5 mmol of diethylzinc (DEZ) was added via syringe and the reaction mixture was heated to reflux for 1 hr. The reaction mixture was then brought to room temperature and the solvent was removed in vacuo to afford the crude solid product. The solid was washed with cold pentane to remove any unreacted amine and carbon disulfide. The products were isolated in yields ranging from 24-65%.

Figure 1: Synthetic scheme for the preparation of ZnO and ZnS precursors.

page 49 NNIN REU 2006 Research Accomplishments • MATERIALS

MATER

IALS

Synthesis of bis(benzoate) zinc: To a round bottom flask was added benzoic acid (5 mmol) and 50 ml toluene. Next was added 4 mmol of diethyl zinc via syringe. The mixture was refluxed for 30 minutes and the solvent was removed in vacuo to afford the product as a white solid in 62% yield.

Nanoparticle Synthesis: Synthesis of the nanoparticles was carried out by adding a trioctylphosphine solution of the appropriate zinc complex to a heated (175°C) solution of trioctylphosphineoxide. The mixture was heated for 2 hrs and then allowed to cool to room temperature. The particles were extracted from solution via the addition of methanol and dispersed in hexane for analysis.


The dialkylamidodithiocarbamate and benzoate based precursors were isolated in yields ranging from 10-62%. The three secondary amines that were used to prepare the dithiocarbamate based metal complexes were N-methylaniline, diethyl amine and diisopropylamine. The lowest yield was obtained from the reaction that made use of N-methylaniline. This is believed to be due to the poor solubility of the amine in pentane. The bis (benzoate) zinc complex was prepared in toluene and isolated in good yield (62.7%). The isolated metal complexes were characterized via FT-IR spectroscopy to confirm the formation of the desired product. The isolated nanoparticles were characterized via UV-Vis spectroscopy and light scattering methods. The expected

λmax for bulk ZnS is 340 nm while that for the particles isolated in our experiments ranged from 272-287 nm. Light scattering data showed that the particles ranged in size from 1184-3279 nm with PDI’s ranging from 0.41-0.69 indicating that the particles were not monodisperse (Figure 3).

The significantly larger than expected (1-20 nm) particle size was due to the fact that the isolated solution was not properly stored after synthesis and prior to analysis. We believe that the particles began to decompose/aggregate as evidenced by the precipitation of solid material prior to the light scattering experiment.

Acknowledgements:

I would like to acknowledge Professors Jason Matthews, Gary Harris and James Griffin for assisting me in completing this project. I would also like to thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for providing me with financial support for the summer research experience.

References:[1] M. Afzaal, K. Ellwood, N.L. Pickett, P. O’Brien, J. Raftery, J.

Waters, Journal of Materials Chemistry 2004, 14, 1310-1315. [2] N.L. Pickett, P. O’Brien, Chemical Record, 2001, 1, 467-479.

Figure 2: Precursor decomposition and nanoparticle formation.

Figure 3: Particle size determination via light scattering.


MAT

ERIA

LS

Abstract:Ripple formation on silicon caused by irradiation with a focused ion beam can create features on the nanometer scale. A phenomenon where a rippled regime nucleates throughout the irradiated area is explored. A model has existed for some time to explain the formation of ripples from ion irradiation. However, there is disagreement between theory and experiment; while theory predicts linear instability at all incident angles, experiments show many cases of flat surfaces under irradiation. Ripple nucleation is a transitional case in which a surface is stable until some perturbation is made.

Nanoscale Materials Morphology using a Focused Ion Beam

Matthew BlosserPhysics, Carleton College

NNIN REU Site: Center for Nanoscale Systems, Harvard UniversityNNIN REU Principal Investigator: Prof. Michael Aziz, Div. Engineering & Applied Sciences, Harvard UniversityContact: [email protected]


MATER

IALS


MAT

ERIA

LS

Abstract:The formation of a diversity of structures via the assembly of colloidal particles is hindered by the scarcity of available particles, the difficulty of attaining mono-dispersed samples, and the lack of tunable and selective interactions. With the aid of depletion interactions, we have been able to induce and control structure in systems of photolithographically designed cylindrical particles. The structures that have been formed thus far consist of isolated columns of cylinders that do not present order on a larger scale. Our aim is to self-assemble these building blocks on a patterned surface to have better control of the structure. Preliminary experiments have shown that the particles adhere to the patterned surfaces as envisioned however, replication of the process is necessary.

Introduction:

The general challenge that is faced by the industry is the fact that at the microscale, researchers typically deal with spherical particles. Armed with a limited diversity of shapes, the result was a limited diversity of structure. We believe that through the use of lithography, we can utilize the diversity of these building blocks and ultimately gain a more mechanicalistic insight into the formation of structure. This project served as a means to answer the question— can we in fact have some form of power on formation and structure? By templating a glass surface, we hoped we could demonstrate selective self-assembly of the particles on a patterned surface.

Self-Assembly of Lithographically-Designed Colloidal Particles on Templated Surfaces

McIntosh BontheraChemical Engineering, New Jersey Institute of Technology

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Prof. Abraham Stroock, Chemical Engineering, Cornell University NNIN REU Mentor: Stephane Badaire, Chemical Engineering, Cornell UniversityContact: [email protected], [email protected]


Experimental Procedure for Glass Coverslips:1. Plasma clean (about 11 glass coverslips

approx. 140 µm in thickness).2. Spin 1 µm layer of SU-8 2002 (6000 rpm, 40

sec).3. Soft bake for 2 min. at 65°C and 3 min. at 95°C

(to remove solvents). 4. Use contact mask aligner (EV620) to expose

pattern onto glass coverslip (exposure time of 5 sec. in top side soft contact setting).

5. Post bake for 1 min. at 65°C and 2 min. at 95°C (to crosslink SU-8 2002).

6. Develop SU-8 2002 by dipping and agitating for 2 min. in SU-8 developer solution, 1 min. in another solution of SU-8 developer solution, and 1 min. in isopropanol solution.

7. Final wash with bottle of isopropanol solution and dry with nitrogen gas.

8. Place in coverslip holder and store.

Surface Chemistry:

We templated a glass surface with shapes made of SU-8. Our goal was to graft an amino silane (positively charged once in water) on the glass surface, and then use

Figure 1

Figure 2


MATER

IALS

this highly positive surface to adsorb a strongly negative polymer, dextran sulfate. The main idea was to make the glass surface strongly negative so that the cylinders would “stick” to SU-8 first, due to the depletion effect. During the surface chemistry trials, it appeared that one coverslip presented a surface activity corresponding to what we were searching for, but we still have to investigate the surface chemistry process to be able to reproduce that result.


The cylinders adhered to the SU-8 patterns like originally dreamt in a nice hexagonal ordered array. Hopefully when placed on a rotating stage, the Brownian cylinders will fall off the glass surface and stack up onto the SU-8 patterns and progressively build themselves. We are currently trying to replicate this phenomenon which is no easy task. It would seem parameters beyond our understanding are involved. This result is a testament that our original goal is attainable and has revitalized our thirst to continue this project and one day build novel structures.

Future Work:

When our coating process is excellent, we plan to explore the consequences of differing concentrations of salts and depletants, confinement effects meaning size and shape variation, and observe if the use of complimentary shapes can affect structure that is formed.

Acknowledgements:

I would like to thank my principal investigator, Professor Abraham Stroock, my mentor, Dr. Stephane Badaire, the Stroock Group for having me, Intel Foundation for sponsoring me, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program (NNIN REU) and National Science Foundation (NSF) for funding, Ms. Melanie-Claire Mallison and Dr. Lynn Rathbun, CNF staff, for their aid, and all who have contributed to this effort.

Figure 4

Figure 3


MAT

ERIA

LS

Abstract:The surface reactivities of pyrite (FeS2) and arsenopyrite (FeAsS) were compared by analyzing how surface characteristics influence the deposition of gold. FeAsS and FeS2 samples were immersed in 100 ppm Au(III) solution for 24 hours. Scanning electron microscopy (SEM) coupled with energy dispersive spectroscopy (EDS) showed that FeAsS had a higher surface coverage of gold after equal exposure time. Surface analysis using x-ray photoelectron spectroscopy (XPS) indicated that gold was reductively adsorbed as Au(0), and that arsenic was oxidized during the reaction in the case of FeAsS. Atomic force microscopy (AFM) was used to image the growth of gold nanoparticles on the FeS2 and FeAsS surfaces as a function of time. The AFM results show preferential growth on surface defects and a higher rate of growth on the FeAsS within the first hour of deposition. In conclusion, the surface of FeAsS was found to better facilitate the reductive adsorption of gold.

Introduction:

Mineral surface reactivity is ultimately dependent on three surface properties: chemical composition, atomic structure (which determines which atoms are exposed to the surface), and microtopography. The purpose of this project was to compare the surface reactivity of FeS2 and FeAsS by analyzing how surface characteristics, such as surface chemistry and topography, influence the deposition of gold.

Gold ore deposits may form in low temperature aqueous environments by reductive adsorption on FeS2 and FeAsS. FeAsS deposits usually contain more gold than FeS2 deposits, and the gold content of arsenian pyrites generally increases with arsenic content. Much of this gold is present as “invisible” gold (particles less than 0.1 µm) [1]. A proposed mechanism for sulfide adsorption involves a redox reaction where gold reduction on As-rich areas is coupled with oxidation at nearby S-rich areas [2].

Understanding the role of Fe, As, S, and Au in these

Nanoparticles in the Environment: A Study of Surface Reactivity of Pyrite and Arsenopyrite

Anthony S. BreitbachChemistry, Clarke College

NNIN REU Site: Michigan Nanofabrication Facility, University of MichiganNNIN REU Principal Investigator: Professor Udo Becker, Department of Geological Sciences, University of MichiganNNIN REU Mentor: Devon Renock, Department of Geological Sciences, University of MichiganContact: [email protected], [email protected], [email protected]

experiments has implications for the recovery of gold from ore deposits, the control of acid mine drainage, and understanding the release of As into the environment.

Gold Adsorption Experiment:

Cleaved FeS2 and FeAsS samples were placed in 100 ppm KAuCl4/1M NaCl for 24 hrs. The FeAsS surface became darker and its solution less yellow, while there was no observable change with the FeS2 sample.

SEM/EDS Analysis:

The exposed samples were analyzed using SEM/EDS. The SEM images in Figure 1 show that FeAsS had a higher surface coverage after 24 hour exposure time compared to FeS2. EDS showed that the adsorbed material contained gold, and the lack of a chlorine peak in the spectra suggested that gold was deposited in a reduced form.

XPS Analysis:

Flat polished samples with a surface roughness of < 0.1 µm were prepared for XPS analysis. XPS is a technique that is able to determine the surface composition and oxidation states of surface components. Here XPS was used to determine how Au(III) is adsorbed on FeS2 and FeAsS by comparing to Au(0) and Au(III) standards. In Figure 2, the FeS2 and FeAsS Au4f peaks are located near the same binding energy as the Au(0) standard, thus Au(III) was reduced on the surface during adsorption.

Figure 1: SEM images of cleaved FeS2 (left) and FeAsS (right) after 24 hours.


MATER

IALS

In addition, the FeS2 peaks exhibit shoulders that extend into the region of the Au(III) standard. This suggests that there may have been adsorption of Au(III) or partially reduced Au(I) on the FeS2 surface. No shoulders on the FeAsS peaks indicate that FeAsS facilitates reductive adsorption to a greater extent compared to FeS2.

In Figure 3, the As3d peaks of the FeAsS standard and the FeAsS sample are compared to determine if As was oxidized and thus facilitating the reduction of gold. The difference in the peak intensities between the standard and the exposed sample indicates a greater proportion of As at a higher oxidation state after exposure. Future studies will determine whether or not the oxidation of As is coupled to the reduction of Au on the FeAsS surface.

AFM Analysis:

Tapping mode AFM was used to image the gold growth on the polished FeS2 and FeAsS surfaces as a function of time. At the beginning of the experiment, both the FeS2 and the FeAsS surfaces appeared relatively smooth.

After 10 minutes, adsorbed material was evident on the surface of both samples, but to a greater extent on FeAsS. By 60 minutes, the FeAsS surface appeared completely covered while the FeS2 surface appeared more sparsely covered.

AFM was also used to see if there was any preferential adsorption due to surface topology. Shown in Figure 4 are the AFM height images of the FeS2 and FeAsS surfaces after exposure to the Au(III) solution for 10 minutes. Both samples displayed preferential adsorption on surface defects as indicated, but to a greater extent on the FeAsS sample.

Conclusions:

It was determined that Au(III) adsorbs to a greater extent on the surface of FeAsS compared to FeS2. Au(III) is reduced to Au(0) on the surface of FeAsS, while the FeS2 surface exhibited signs of Au(III), Au(0), and partially reduced Au(I). The oxidation of As during Au(III) adsorption on FeAsS is one possible explanation for the greater extent of reduction. It was also found that surface defects help assist Au(III) adsorption. In conclusion, the surface of FeAsS was found to better facilitate the reductive adsorption of gold.

Acknowledgments:

The Computational Mineralogy & Surface Science Group within the Department of Geological Sciences at the University of Michigan, Devon Renock and Professor Udo Becker, and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Maddox, L.M. et al. (1998) Invisible gold: Comparison of Au

deposition on pyrite and arsenopyrite. American Mineralogist, 83, 1240-1245.

[2] Schaufuss, Andrea G. et al. (2000) Reactivity of surface sites on fractured arsenopyrite (FeAsS) toward oxygen. American Mineralogist, 85, 1754-1766.

Figure 2: XPS Au4f peak comparison between Au(0) and Au(III) standards, and exposed FeS2 and FeAsS samples.

Figure 3: XPS As3d peak comparison between FeAsS standard (top) and exposed FeAsS (bottom).

Figure 4: 20 µm x 20 µm AFM images of polished FeS2 (left) and FeAsS (right) after 10 minute exposure.


MAT

ERIA

LS

Abstract:For the microelectronics industry to begin manufacturing smaller and faster devices, thin films utilized as interconnect structures must have lower dielectric constants and increased thermo-mechanical stability. The introduction of nanoporosity is often used to fine-tune electrical properties of certain materials. Oxycarbosilane (OCS), an organosilicate spin-on glass, has shown not only an extremely low dielectric constant, but has also exhibited a high level of mechanical reliability in comparison to similar nanoporous thin film glasses. During integrated circuit fabrication, thin films are subjected to active chemical solutions which can diffuse into the material, adversely affecting device performance and reliability. The current study focuses on the extent to which different solutions diffuse into OCS thin films with varying porosity levels and pore size characteristics.

Introduction:

Two forms of OCS, a bridged precursor molecule, were studied. These variations of OCS, namely ethylene and methylene, are defined by the specific organic group

Diffusion of Aqueous Solutions in Oxycarbosilane Nanoporous Thin Films during Processing of Interconnect Structures

Sarah BryanPhysics, Florida International University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford UniversityNNIN REU Principal Investigator: Prof. Reinhold H. Dauskardt, Materials Science & Engineering, Stanford UniversityNNIN REU Mentor: David Maxwell Gage, Department of Materials Science & Engineering, Stanford UniversityContact: [email protected], [email protected]

acting as the bridging bond. Nanopores were created by introducing sacrificial porogen molecules coupled with post-deposition annealing to remove porogen remnants, leaving behind pores of various sizes. Previous x-ray diffraction tests have revealed an organized, symmetric structure of nanopores in certain OCS thin films and a completely random distribution of pores in others.

Organic groups present in organosilicate glasses typically result in highly hydrophobic materials. Experiments with similar nanoporous organosilicates have shown rapid diffusion in the presence of buffering solutions. For this reason, initial testing of OCS thin films was performed in concentrated buffer solutions of varying pH.

Procedure:

The OCS films were applied to silicon wafers using a spin-casting technique, resulting in film thicknesses of 500 nm. Various samples included two forms of OCS, three pore sizes, and two porosity percentages. All samples were capped with a 200 nm transparent layer of silicon nitride using a plasma enhanced chemical vapor deposition system.

Samples were cleaved into approximately 50 mm squares immediately before immersing the films into solution. Deionized water and two buffer concentrates of pH 3 and pH 11 were tested. Photomicrographs were taken with an optical microscope at 5x magnification [Figure 1] in order to measure diffusion distance. All data was analyzed using Fick’s Law of Diffusion which represents a linear relationship between diffusion distance and the square root of time.

Results:

As illustrated in Figures 2 and 3, two general trends were confirmed by experiment. With regard to both forms of OCS, methylene and ethylene, smaller pores present a better barrier to diffusion. Figure 2 shows a steady decline in the diffusion coefficient as pore radius

Figure 1: Photomicrograph of sample in solution showing a sharp diffusion front.


MATER

IALS

decreases. Correspondingly, samples with the lowest porosity percentage exhibited the slowest diffusion rates, as depicted in Figure 3. In addition to the differences in porosity percentage, Figure 3 also demonstrates the relationship between the three solutions for a specific OCS thin film. The graph shows that porosity percentage has a much larger impact on diffusion rates than the solution. When comparing data for ethylene and methylene, as in Figure 4, all thin films and solutions showed faster diffusion rates in ethylene, with the exception of two methylene films. When considering nanopore organization alongside the data collected, nearly all ethylene samples displayed a “not organized” structure of nanopores, while most methylene thin films showed signs of organization. The two methylene films described as “not organized” were the same two films that exhibited the only occurrence of higher diffusion rates in methylene when compared to ethylene.

Additionally, an organosilicate glass similar in total composition and overall porosity created with a chemical vapor deposition (CVD) process was subjected to identical diffusion experiments. The main difference between the CVD glass and OCS was the placement of the organic groups, which are found as terminal bonds instead of bridging bonds in the CVD glass. No diffusion was observed for the samples tested.

Conclusion:

Although oxycarbosilane nanoporous thin films are ultra low-κ materials with a high level of mechanical reliability, OCS displays extremely low resistance to diffusion of any aqueous solution tested. This process seems to be further enhanced by a lack of nanopore organization. If experimentation could be performed on OCS thin films exhibiting different organization

levels, the mechanism responsible for alignment of the nanopores may be isolated and understood. This could result in future materials being manufactured with highly symmetric porous structures which may ultimately lead to drastically reduced diffusion rates. With respect to the lack of diffusion seen in the CVD glass, the placement of the organic groups obviously lends to the overall hydrophobicity of the material. Such a drastic difference in diffusion rates between the two organosilicate glasses may also mean that the organic groups used in OCS are possibly of more importance than other factors mentioned.

Because of the large amount of different OCS thin films and solutions used in this project, time permitted only one series of experiments for each specimen and testing condition. To be sure of the general trends discussed, these diffusion experiments should be repeated to ensure reproducibility.

Acknowledgements:

I’d like to thank Professor Reinhold H. Dauskardt and David Maxwell Gage for welcoming me into their group and providing answers to all my questions. I’d also like to a give special thank you to Mike Deal and Maureen Baran, as well as the rest of the staff at the Stanford Nanofabrication Facility. Additionally, thank you to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for providing such an amazing opportunity.

References:[1] D. Geraud, T. Magbitang, W. Volksen, E. E. Simonyi, R.

D. Miller, “New Spin-On Oxycarbosilane Low-k Dielectric Materials With Exceptional Mechanical Properties,” 2005 International Interconnect Technology Conf, Burlingame, CA.

Figure 2: Comparison between different pore sizes of a specific film and solution.

Figure 3: Comparison between porosity percentages of a given OCS film.

Figure 4: Rare occurrence in which methylene showed faster diffusion rates than an identical ethylene film.


MAT

ERIA

LS

Abstract:The thermal decomposition of iron(Fe)-bearing carbonates to form magnetite nanoparticles was studied in order to better understand the parameters that control nanoparticle formation. Rates of decomposition were determined for natural siderite (FeCO3) and ankerite (Ca(Mg,Fe,Mn)(CO3)2). An activation energy of 153 kJ/mol was determined for the decarbonation of siderite. Characterization of the magnetite nanoparticles using TEM, SEM, and EDS analysis showed a progressive increase in particle size and Mg/Mn contents as the reaction proceeded.

Introduction:The origin of nanophase magnetite in Martian meteorite

ALH84001 has fueled the debate as to whether there was ancient bacterial life on Mars. McKay [1] suggested that these 10-100 nm magnetite particles shared many of the characteristics of magnetite produced by magnetotactic bacteria on Earth, indicating a possible biogenic origin. An alternative, inorganic hypothesis (Brearley [2]) proposes thermal decomposition of Fe-bearing carbonates. To further explore the thermal decomposition mechanism, heating experiments have been carried out on siderite and ankerite to determine the reaction rates and characteristics of the products as a function of temperature and heating time.

Methods:Heating experiments were performed by placing about

20 mg of uncrushed carbonate sample grains in a gold foil packet which was suspended in a Deltech furnace at the desired temperature and at 1 atm in a CO2 environment. Reaction progress was measured by weight loss after completion of the experiment.

The starting materials and run products were characterized by x-ray diffraction (XRD), electron microprobe (JEOL 8200), scanning electron microscopy (SEM; JEOL 5800LV, Hitachi S-5200), transmission electron microscopy (TEM; JEOL 2010F) combined with energy dispersive spectroscopy (EDS). The percent Fe content (wt% Fe) of the ankerite determined by electron microprobe, was used to calculate maximum and minimum theoretical weight losses, assuming completed decomposition of the siderite component of the carbonate.

Formation of Magnetite Nanoparticles by Thermal Decomposition of Iron Bearing Carbonates: Implications for the Evidence of Fossil Life on Mars

Alicia CohnChemistry, Gettysburg College

NNIN REU Site: Nanoscience @ UNM, University of New MexicoNNIN REU Principal Investigator: Dr. Adrian J. Brearley, Department of Earth and Planetary Sciences, University of New MexicoNNIN REU Mentor: Jana Berlin, Department of Earth and Planetary Sciences, University of New MexicoContact: [email protected], [email protected]

Results and Discussion:The decomposition of siderite occurred in 30 seconds to

2.5 hours in the temperature range of 500°C to 700°C. The data followed a sigmoidal curve and were fitted to the Avrami-Erofeev rate equation. The rate constant at each temperature was determined to give an activation energy of 153 kJ/mol, comparable to the literature value of 183 kJ/mol for the decomposition of natural siderite [3].

The decomposition of ankerite was examined between 600°C and 800°C. At 600°C, weight loss did not change over time but at 650°C, significant weight loss occurred and the samples became magnetic. The predicted weight loss for the reaction was, however, only reached after 40 minutes at 700°C and 1 minute at 800°C.

For magnetite nanoparticles formed in the 700°C experiments, grain size was measured from TEM images. (Figure 1.) Thirty different grains were measured across the longest axis from each experiment. Grain size and distribution increased with heating duration to 20 min, but shrank after 40 min (Figure 2). Magnetite grain sizes were also measured from samples run for 20 min and 180 min at 650°C and have grain sizes and distributions comparable to those of the 700°C set of experiments.

Figure 1: HRTEM image of magnetite reaction product (ankerite: 40 min, 700°C).


MATER

IALS

Magnetite compositions were determined by EDS analysis for samples from the 700°C series. The data show a progressive increase of Mg and Mn contents of magnetite with time, but reach a plateau after 20 min (Figure 3). This increase in Mg and Mn contents is probably the result of the progressive, but slower decomposition of Mg and Mn carbonates in the ankerite. The surface of one reacted ankerite (700°C for 40 min) was examined by FEGSEM. The surface appeared much rougher and had higher porosity compared to the smooth carbonate surfaces observed before heating (Figure 4).

Conclusion and Future Research:The magnetite nanoparticles produced in these experiments

support some aspects of the inorganic hypothesis, but there are still many unanswered questions. This study shows that thermal decomposition can reliably produce nanophase magnetite particles with grain sizes consistently < 100 nm, the maximum size of the nanoparticles observed by McKay in ALH84001 [1].

The magnetite found in ALH84001 and produced by bacteria is pure Fe3O4. However, in these experiments the magnetites contain significant amounts of Mn and Mg, whose

Figure 4: Cluster of magnetite particles on reacted ankerite. Secondary electron image (FE-SEM).

Figure 2: Magnetite grain size as a function of time at 700°C.

concentrations increase in time. Nevertheless, the shortest ankerite heating experiments formed magnetite particles with high weight percent Fe contents. This result indicates that Fe-rich particles might be produced when complex carbonate decomposes rapidly such as during a shock event.

Our SEM observations revealed magnetite grains on the surface of the carbonate rather than embedded in the carbonate as occurs in the meteorite. Further TEM studies of cross sectional samples from carbonate grains are needed to establish whether magnetite has developed within the interior of the carbonate. In addition to variations of temperature and duration, more experiments should be done at different pressures to establish the possible effects of shock on the decomposition reaction.

Acknowledgements:Dr. Adrian Brearley, Jana Berlin, Dr. Rhian Jones, Dr. Ying-

Bing Jiang, Dr. Peng Li, Jim Connolly, UNM REU, National Nanotechnology Infrastructure Network REU Program, and National Science Foundation.

References:[1] McKay, D.S., Gibson E. K. Jr., Thomas-Keprta K. L., Vali H.,

Romanek C. S., Clemett S. J., Chillier X. D. F., Maechling C.R., and Zare R.N (1996), “Search for past life on Mars: Possible relic biogenic activity in Martian meteorite ALH84001” Science 273, 924-930.

[2] Brearley A. J. (1998) “Magnetite in ALH84001: Product of decomposition of ferroan carbonate.” Lunar and Planetary Science XXIX, Abstract no. 1451, CD-ROM.

[3] Gotor F.J., Macias M., Ortega A., and Criado J. M. (2000) “Comparative study of the kinetics of the thermal decompostion of synthetic and natural siderite samples”; Physics and Chemistry of Minerals 27, 495-503.

[4] Treiman, A.H. (2003) “Submicron magnetite grains and carbon compounds in Martian meteorite ALH83001: Inorganic, abiotic formation by shock and thermal metamorphism” Astrobiology 3, 369-392.

Figure 3: Fe-, Mg-, and Mn- % weight content of magnetite as a function of heating duration (ankerite, 700°C).


MAT

ERIA

LS

Abstract:In recent years, one-dimensional solids like nanowires have received increased attention as building blocks for future nanoscale devices. Their small length scale and unique physical and chemical properties make them interesting materials from technological and pedantic viewpoints. In order to fully exploit their advantages, it is necessary to perform experimental characterization on single nanowires. As a first step towards this goal, we are aiming to synthesize zinc oxide nanowires at the site of experimentation. The final goal of this research is to perform mechanical and electromechanical characterization experiments on the nanowires by integrating the nanowire synthesis process with micro-electromechanical systems (MEMS) experimental test-bed. The nanowires are grown by vapor-liquid-solid (VLS) mechanism. The zinc oxide nanowires are synthesized using 1:1 weight mixture of zinc oxide powder and graphite as precursors and gold as catalyst. At 1000°C the graphite reduces zinc oxide to form zinc and oxygen vapor. The nanowire growth is initiated when the gold (evaporated on (100) silicon substrate) is saturated with zinc vapor. The diameters of the nanowires are between 30 nm to 200 nm and the lengths are up to 20 µm.

Introduction:

When the size of material decreases (to micro and nanoscale) many interesting phenomena begin to emerge. Since the dimensions of nanowires coincide with the critical material length scales, they are expected to exhibit unique properties. As a result, they will serve as

Site Specific Nanowire Growth

Sonia Y. Cortes-JimenezMechanical Engineering, University of Puerto Rico - Mayaguez Campus

NNIN REU Site: Penn State Nanofabrication Facility, The Pennsylvania State UniversityNNIN REU Principal Investigator: Dr. Aman Haque, Nuclear and Mechanical Engineering, The Pennsylvania State UniversityNNIN REU Mentor: Amit Desai, Nuclear and Mechanical Engineering, The Pennsylvania State UniversityContact: [email protected], [email protected]

an excellent system to investigate mechanics at different length scales, coupling between mechanical, electrical and thermal properties and their effect on dimensionality and size reduction. Their mechanical properties coupled with quasi-one-dimensional nature of electronic states make the nanowires potential material for future nanoscale sensors and actuators. Nanowires will be integral in the reduction of size of functional systems and in the enhancement of performance of sensors and actuators. However in order to fully exploit their advantages, it is necessary to perform experimental characterization on single nanowires. One of the challenges is the ability to have the specimens at the site of experimentation. We propose to address this issue by synthesizing zinc oxide nanowires at the site of experimentation. We chose microelectromechanical systems (MEMS) devices for experimental characterization because of the ability to perform in situ uniaxial experiments. We chose zinc oxide because ZnO is a key technological material; it is a semiconductor as well as a piezoelectric material. Zinc oxide is also bio-safe and biocompatible.

The main objective of this research was to study site specific growth of zinc oxide nanowires. We also wanted to study the effects of different process parameters on nanowire growth and control the growth sites and morphology of the wires.


In this research we grew the zinc oxide nanowires via the VLS mechanism using gold as a catalyst. We started with zinc oxide powder and graphite powder in a 1:1 ratio in an alumina crucible inside the furnace. We placed the silicon samples (100) with 20 nm gold patterns on them downstream from the crucible. As the temperature of the crucible increased to ~ 1000°C, the ZnO powder was reduced by graphite to form zinc and CO vapors. The corresponding chemical reaction can be expressed as:

ZnO(s) + C(s) ‡ Zn(g) + CO(g)CO(g) + ZnO(s) ‡ CO2(g) + Zn(g)

The argon gas carried the zinc, CO and CO2 vapors to the samples. Meanwhile the formation of eutectic

Figure 1: VSL schematic.


MATER

IALSalloy droplet occurred at each catalyst site. The gaseous

products produced by the above reactions would adsorb and condense on the alloy droplets. Subsequently, the following reaction was catalyzed by the Au-Si alloy at solid- liquid interface to obtain zinc oxide nanowires.

Zn(g) + CO(g) ‡ ZnO(nanowire) + C(s)C(g) + CO2(g) ‡ 2CO(g)

The zinc vapor saturated the alloy droplet, followed by the nucleation and growth of solid ZnO nanowire due to the super saturation of the liquid droplet. Incremental growth of the nanowire taking place at the droplet interface constantly pushes the catalyst upwards. Thus, such a growth method inherently provides site-specific nucleation at each catalytic site.


From the existing scanning electron microscope (SEM) images, we estimated the nanowire diameters to be in a range from 30 nm to 200 nm and the lengths up to 20 µm. In some of the nanowires we observed a gold tip at the end of the nanowire providing evidence for VLS mechanism-based growth. In the first run, which was set at 10 sccm argon flow rate, we obtained nanowire growth from 500°C to 910°C. During the second run, also set at 10 sccm, growth appeared from 500°C to 800°C. We also observed some nanobelt growth in the lower temperature region.

We observed no growth at higher flow rates and this is mainly due to the zinc vapors being carried away by the argon leading to low probability for nanowire growth reactions. We also experimented with a different substrate; we tried to grow nanowires on silicon dioxide (SiO2) at 10 sccm; we did not obtain any growth and this is mainly because SiO2 is an amorphous substrate.

Future Work:

We will fabricate MEMS devices out of silicon. We will pattern gold on the devices on the desired area and we will grow the zinc oxide nanowires with gold as catalyst. Then we will perform uniaxial tensile experiments on the nanowires inside a SEM using the MEMS device. From the experiments we will estimate the mechanical quantities like stress and strain and electrical entities like voltage and resistance. We will use this information to study the mechanical and electro-mechanical (piezoelectric) properties of nanowires.

Conclusion:

We synthesized ZnO nanowires on a Si substrate by VLS growth process using Au as catalyst. We observed no growth at higher flow rates or on the amorphous substrate. There was furnace tube degradation at every growth run. We also observed nanobelt formation at lower temperatures.

Acknowledgements:

National Science Foundation, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and The Pennsylvania State University. Special thanks to Dr. Aman Haque and Amit Desai from the Department of Mechanical and Nuclear Engineering. Thanks to Andrzej Mieckowski, John McIntosh and the staff of the Nanofab Facility at PSU.

References:[1] Yang, Peidong et al. “Controlled Growth of ZnO Nanowires

and Their Optical Properties.” Advanced Functional Materials 12 (2002): 323-331.

[2] Desai, Amit V. ; Haque, Aman M. “Mechanical Testing of ZnO Nanowires.” Comprehensive Report. PA. 2005.

Figure 2: Nanowire at 910˚C.


MAT

ERIA

LS

Abstract:Available forms of silicon carbide exist as 6H-SiC and 4H-SiC polytypes, but suffer from crystal defects such as micropipes, which affect the electrical properties of the material. Because of its advantageous properties, lack of micropipe defects, equivalent high electron and hole mobilities, and non-anisotropic nature, cubic silicon carbide (3C-SiC), also called β-SiC, has potential applications in semiconductors and microelectronics. In this project, the production of high quality β-SiC, is grown on an AlN-on-SiC substrate by chemical vapor deposition. The growth of SiC was successful, but we were unable to characterize the growth.

Introduction:

Due to silicon’s inability to work at high temperatures and the increased need for semiconductor materials to withstand harsh environments, alternatives to silicon are becoming increasingly popular. Silicon carbide (SiC), over the past several years, has become popular as a semiconductor material to meet these requirements. Some of its properties, which include a wide bandgap, high saturated electron drift velocity, high electric breakdown field, high thermal conductivity, high bonding energy, and resistance to oxidation, creep, and corrosion, make SiC very useful. This is especially true when used in high frequency, high temperature, and high power devices, as well as in harsh and high radioactive environments. SiC has over 200 different polytypes, but the most commonly known forms are 3C, 4H, and 6H. Production of SiC mostly exist as 4H and 6H polytype forms, but these polytypes suffer from micropipe defects, which hampers the electrical properties of these polytypes. 3C-SiC, or β-SiC, properties, which include the lack of micropipe defects, a non-anisotropic nature, and equivalent high electron and hole mobility, give it an advantage over the 4H and 6H polytypes. 3C-SiC can be used in CMOS electronics and in field effect transistors (FET's), but are thermodynamically unstable. Growth of 3C-SiC is a major developmental challenge, given that most SiC wafers are 4H and 6H. Growth methodologies

β-SiC Growth on AlN-on-SiC

Henry Daise IIIComputer Science, Morehouse College

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard UniversityNNIN REU Principal Investigator: Dr. Gary L. Harris, Director,

Howard Nanoscale Science & Engineering Facility, Howard UniversityNNIN REU Mentor: Mr. Crawford Taylor, Howard Nanoscale Science & Engineering Facility, Howard UniversityContact: [email protected], [email protected], [email protected]

for β-SiC growth include seed-sublimation, electron cyclotron resistance, and epitaxial growth methods such as and liquid phase epitaxy and vapor phase deposition [3,4]. In the literature, the growth of 3C-SiC has been done on silicon substrates but the quality of the growth has led to discovering voids and other crystal defects [2]. This project's aim is to produce high quality β-SiC on an AlN-on-SiC substrate via chemical vapor deposition (CVD).


Growth of silicon carbide was done in a horizontal cold wall CVD reactor. The first group of trials were done using silicon wafers of <100> and <111> orientations, and the second using Si <100>, 6H-SiC, and AlN-on-SiC wafer substrates. All trials used 8 slpm of H2 as the carrier gas and a chamber pressure of 200 Torr. For our trials of SiC on silicon <100> and <111>, a buffer layer was grown using 20 sccm of propane for 2 minutes. After the buffer layer was grown on the substrate, the reactor was ramped up to its growth temperature, which ranged from 1050-1300°C, and growth times were 10 and 30 minutes. The second set of trials used 120 sccm

Figure 1: Silicon carbide growth on Si <111> substrate.


MATER

IALS

of hydrogen chloride (HCl) gas, for the in situ etching of the oxide layers on our substrates, for 10 minutes at temperatures from 1000-1020°C. Growth temperatures for the three substrates ranged from 1160-1305°C, and the growth time was 30 minutes.

Results:

After growth of the silicon carbide on Si <100> and Si <111>, thickness measurements were taken for each trial using an ellipsometer. Growth measurements indicated thicknesses up to 200 nm. We were unable to obtain thickness measurements for the 6H-SiC, and AlN-on-SiC group, because the refractive indexes of both were close. Surface morphology measurements of the substrate and epilayer growths were done by atomic force microscopy (AFM). For the Si group, the crystal growth was moderate and it took on the orientation of the substrate, as seen in Figure 1 and 3. In the AlN-on-SiC group, the HCl etching decreased the substrates surface roughness, before growth. When the SiC/AlN-on-SiC was observed, as seen in Figures 2 and 3, at a temperature of 1160°C, crystal growth shows clearly defined geometrical shapes as well as a tendency for the substrate crystal structure to be near the <111> orientation of Si. As the growth temperature increased, the crystal grain size tended to decrease, except for layers grown at 1305°C.

Conclusions/Future Work:

Growth of SiC on AlN-on-SiC substrate was successful, but we were not able to characterize β-SiC on

the substrate. When observed by AFM, the crystal grain size decreased, as the temperature increased. Future work includes determining growth rates with and without HCl etching, and comparing the electrical characteristics of 3C-SiC, when grown on different substrates.

Acknowledgements:

I would like to thank Mr. Crawford Taylor, Dr. Gary L. Harris, Mr. James Griffin, Dr. Peizhen Zhou, the staff and students at HNF, the National Science Foundation (NSF), and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program (NNIN REU).

References:[1] Birudavolu, S., Luong, S. Q., Nuntawong, N., Xin, Y. C., Hains,

C. P., & Huffaker, D. L. (2005). In-situ mask removal in selective area epitaxy using metal organic chemical vapor deposition. Journal of Crystal Growth, Vol. 277, 97-103.

[2] Gupta, A., Paramanik, D., Varma, S., & Jacob, C. (Oct. 2004). CVD growth and characterization of 3C-SiC thin films. Bull. Mater. Sci., Vol. 27, No. 5, 445-451.

[3] Müller, G., Krötz, G., Niemann, E. (1994).SiC for Sensors and High Temperature Electronics. Sensors and Actuators A. Vol. 43, 259-268.

[4] MRS Bulletin, (March 1997) Silicon carbide electronic materials and devices, Vol. 22, No. 3.

[5] Polychroniadis, E. K., Andreadou, A., & Mantzari, A. (March 2004). Some Recent Progress in 3C-SiC Growth a TEM Characterization. Journal of Optoelectronics and Advanced Materials. Vol. 6, No. 1, 47-52.

Figure 2: Silicon carbide growth on AlN-on-SiC at 1160°C. Figure 3: Silicon carbide growth on AlN-on-SiC at 1250°C.


MAT

ERIA

LS

Abstract:Rational assembly of “bottom-up” nanoscale structures such as nanowires has become a bottleneck limiting their real-world applications. Recent studies have achieved orientation control of the nanowires via directed assembly by fluidic flow or field-assisted techniques. However, these approaches still lack the control over location and end-to-end registry. In this project, we study rational assembly of nanowires with precise orientation and location control through dielectrophoresis (DEP). DEP assembly relies on the interaction of the induced dipole moment and an inhomogeneous electric field. As a result, the motion of a neutral subject, such as a nanowire, can be controlled by external voltages on nearby electrodes, leading to precise deposition of nanowires across the electrodes. The first part of this report discusses design and fabrication of the electrode patterns with features ranging from 0.5 µm to 7.0 µm, which enable the investigation of nanowire alignment with respect to feature size and nanowire length. In the second part, we report studies to explore the effects of external parameters such as voltage and frequency on DEP assembly. We report device yield over 80% via DEP.

Introduction:

As integrated circuits continue to aggressively scale down in size, novel nanostructures such as nanowires (NWs) have attracted increasing interest as essential components in future electronics. To date, the most widely used NW assembly techniques, such as directed assembly by fluidic flows [1], do not address the primary challenge of acquiring control of the position and registration of “bottom-up” grown NWs with the other components in a nano-scale device. DEP offers a potential solution to these problems and also holds the advantage of compatibility with established integrated circuit production [2]. In this project, DEP is used to precisely position NWs, utilizing the NW’s high-aspect ratio and the tendency for matter to polarize in the presence of strong inhomogeneous electric fields and migrate to the region of highest field intensity [3]. By

Rational Assembly of Semiconductor Nanowires via Dielectrophoresis

Ying Yi DangApplied Physics, Columbia University

NNIN REU Site: Michigan Nanofabrication Facility, University of MichiganNNIN REU Principal Investigator: Dr. Wei Lu, Electrical Eng. & Computer Science / Applied Physics, University of MichiganNNIN REU Mentor: Wayne Y. Fung, Electrical Eng. & Computer Science, University of MichiganContact: [email protected], [email protected]

creating a photolithography-defined microscale electrode pattern that can produce an electric field that varies in a controlled manner across the sample, the effects of voltage, frequency, and electrode spacing on DEP can be studied and optimized.


A reticle that allows for simultaneous studying of multiple electrode spacings was created and designed with pairs of long, rectangular “bus” electrodes spaced 5, 15, 25, 40, and 50 µm apart, featuring evenly spaced pairs of “floating” electrodes of 1 x 4 µm in size and distanced 0.5 to 7.0 µm from the left and right bus electrodes (Figure 1). The floating leads design ensured parallel production of individually addressable nanowire devices. It also helped prevent burnout once a NW was bridged, as would normally occur from the heat generated by the AC signal during dielectrophoresis [4]. Silicon wafers with 500 nm of oxide were spin coated with LOR3A liftoff resist and SPR220-3.0 photoresist, exposed with a projection lithography system, developed with MIF300,

Figure 1: Reticle pattern for DEP testing.


MATER

IALS

evaporated with 50 nm of Ti and 600 nm of Au, then placed in liftoff solution to remove excess metal.

DEP was then carried out to deposit NWs across the bus and floating leads. The applied voltage values ranged from 1.0-20.0 volts and frequencies were varied from 50 kHz to 1 MHz. In each DEP trial, a 0.5 µL drop of heavily-doped p+Si NW solution was dispensed onto the sample, and the AC signal was turned on for 50 s.


The effect of voltage on NW alignment was evident (Figure 2). As voltage was increased, more NWs were drawn towards the electrode pairs, and few NWs were observed outside the active device region because NWs had all migrated towards the regions of high field intensity (Figure 3). This was consistent with the expectation that higher voltages would create stronger electric fields, which in turn would draw in more NWs. NW alignment yields of up to 80% could be achieved. However, if the applied voltages were excessively high, it could render insignificant the field distorting effect

of the floating fingers, causing NWs to densely align as though the floating leads were absent.

The device yield showed a non-monotonic dependence on frequency (Figure 4). The drop off of yield at very high frequencies was likely due to the inability of the nanowire polarization to follow up with the applied field. However, further studies including the dependence of frequency on DEP with respect to the finger-bus spacing will be needed to obtain a complete picture.

Future Work:

Further study of DEP would require research on various types of NWs and nanoparticles to explore other factors such as particle geometry and relative polarizabilities. It would also be possible to analyze effects of electrode patterning by designing floating leads with sharper ends mirrored by similarly sharp teeth on the surface of the electrodes. Such a design may further enhance the DEP effect and increase alignment precision.

Acknowledgments:

I would like to extend many thanks to Dr. Wei Lu, Wayne Y. Fung and the rest of the Lu Group for their research support, to Sandrine Martin, Deb Swartz and the SSEL staff for their facilitation of my experience at MNF, and to National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for funding.

References:[1] Y. Huang, et al., Q. Wei, C.Lieber, Science 291, 630 (2001).[2] S. Evoy, et al., MEE 75, 31 (2004).[3] H.A. Pohl, J. Appl. Phys, 22, 869 (1951).[4] C.S. Lao, et al., Nano Lett. 6, 263 (2006).

Figure 2: NW alignment yield for 1 µm finger-electrode spacing.

Figure 3: Optical image of NWs between fingers and electrodes. Figure 4: Effect of frequency on NW alignment yield.


MAT

ERIA

LS

Abstract:Bulk gold (Au) is diamagnetic but 2 nm dodecanethiol (thiol) capped gold nanoparticles (NPs) have been reported to exhibit ferromagnetism. This ferromagnetism is believed to result from spin-orbit coupling between surface-bound thiols and gold surface atoms. As the gold nanoparticle size decreases and the surface area to volume ratio increases, the likelihood of ferromagnetism increases.

The size dependence of the magnetic properties of thiol-capped gold nanocrystals was studied. Thiol-capped gold nanocrystals ranging from 2 to 6 nm in diameter were synthesized using colloidal methods. Their magnetic susceptibilities were measured using a superconducting quantum interference device (SQUID) at room temperature and 5 Kelvin (K). Contrary to two published reports, but consistent with another study, the thiol-coated gold nanocrystals did not exhibit ferromagnetism, even with diameters of 1.8 nm.

Introduction:

The high surface area-to-volume ratio in nanocrystals can lead to unexpected properties. Of interest here is the past observation that Au can become paramagnetic—and even ferromagnetic under some conditions-when the diameter is small enough—even though it is diamagnetic

Measuring the Size Dependence of the Magnetic Properties of Alkanethiol-Coated Gold Nanocrystals

Sarah C. HernandezPhysics and Astronomy, Texas Christian University

NNIN REU Site: Microelectronics Research Center, The University of Texas at Austin NNIN REU Principal Investigator: Dr. Brian A. Korgel, Chemical Engineering, The University of Texas at Austin NNIN REU Mentor: Andrew Heitsch, Chemical Engineering, The University of Texas at Austin Contact: [email protected], [email protected]

as a bulk material. Diamagnetic susceptibilities are small and negative, whereas paramagnetic materials have small positive susceptibilities. Ferromagnets have large positive susceptibilities and below their Curie temperature spontaneously magnetize. The objective of this research project was to determine the size dependence of the magnetic susceptibility of Au NPs and how capping ligand chemistry influences their magnetic properties. Thiol-capped Au NPs were synthesized with 2 nm and 6 nm average diameters. Au NPs were also synthesized with tetraoctylammonium bromide (TOAB), polyvinyl pyrolidone (PVP), and polyally amine hydrocloride (PAAHC) as capping ligands. The magnetic properties of the nanocrystals were measured at 5K and room temperature.


2 nm thiol-capped gold NPs were prepared by arrested precipitation [2,7]. An aqueous solution (15 mL, 0.03M) of hydrogen tetrachloroaurate-hydrate, and an organic solution of TOAB (40 mL, 0.12M) in toluene were prepared, combined, and stirred vigorously for 30 minutes. TOAB complexes with the gold salt and transfers it to the organic solution. After 30 minutes of stirring the two phases were separated and the organic phase was retained. Dodecanethiol (0.4028 mL) was added to the organic solution. A freshly prepared aqueous reducing solution of sodium borohydride (12.5 mL, 0.40M) was mixed with the organic solution and stirred vigorously for 2 hours. The NPs were isolated by precipitation with methanol followed by centrifugation at 10°C and 8000 rpm for 10 minutes. By TEM, the NPs had a diameter of 1.8 ± 0.26 nm.

Larger diameter (6 nm) Au NPs were synthesized by reducing the TOAB:Au salt complex prior to adding dodecanethiol. TOAB serves as the capping ligand, but its weaker bonding with the Au surface enables larger particles to be obtained. 24 hrs after adding the sodium borohydride reducing solution (30 mL, 0.40 M), 240 µL of dodecanethiol was added. These NPS were

Figure 1: Hi-Res TEM image of 1.8 ± 0.26 nm Au Nps.

Figure 2: Low-Res TEM image of 5.5 ± 0.62 nm Au Nps.


MATER

IALS

precipitated with methanol and isolated by centrifugation. According to TEM, the Au NPs had an average diameter of 5.5 ± 0.62 nm.

SQUID Sample Preparation:

Kapton® tape was placed adhesive side up and Au NPs were drop-cast onto the tape and dried. The Kapton® tape with the sample was then wrapped around a quartz tube, sandwiching the sample between the two. The quartz tube was inserted into a straw to measure the magnetic response of the Au NPs.

Quartz Sample Blank:

To ensure accurate magnetic measurements, the background—including the quartz tube and Kapton® tape—was measured and subtracted. The scanned length of the quartz tubing was ~ 4 cm with a weight of ~ 0.64g. The magnetic susceptibility of of bulk quartz is χtabulated = -1.03 x 10-9 emu/gOe. Copper(II) oxide (CuO) was used as a standard to check that the sample was properly cantered and that the instrument was accurately calibrated. The magnetic susceptibility of CuO is χbulk = 2.99 x 10-6 emu/gOe. Measurements of CuO showed a paramagnetic response comparable to the literature value.


Both 1.8 nm and 5.5 nm diameter thiol capped Au NPs exhibited diamagnetic responses at 5K and 300K with magnetic susceptibilities of χ5K = -1 x 10-7 emu/gOe, χ300K = -9 x 10-7 emu/gOe, χ5K = -3 x 10-7 emu/gOe and χ300K = -6 x 10-7 emu/gOe, which is in good agreement with bulk Au, χbulk = -1 x 10-7 emu/gOe. To ensure that there was a measurable signal from the sample above the background, 13.8 mg of the 1.8 nm diameter NPs and 9.62 mg of the 5.5 nm diameter NPs were measured. These data appear to indicate that dodecanthiolcapped Au NPs are diamagnetic, regardless of size. Further research, however, is needed to confirm whether this is indeed the case. PVP-capped Au NPs were also synthesized, but their magnetic response has not yet been measured.

In conclusion, Au NPs were synthesized using colloidal methods and their magnetic properties were measured at 5K and 300K using SQUID magnetometry. The thiol-coated Au nanocrystals appear to be diamagnetic down to diameters of 1.8 nm. No evidence of ferromagnetism at temperatures 5K and higher was observed in any samples.

Future Work:

Future work includes exploring variations of Au NP capping ligand chemistry. PVP has been reported to give rise to larger paramagnetic signals than the thiols, for example. Transition metals, like Pt, are particularly interesting for future studies and a comparison to Au.

Acknowledgements:

Acknowledgements are extended out to Dr. Brian A. Korgel and the Korgel research group, Andrew Heitsch, NSF, and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Crespo, P., et al. “Permanent Magnetism, Magnetic Anisotropy,

and Hysteresis of Thiol-Capped Gold Nanoparticles,” Phys. Rev. Lett. 93 (2004) 087204.

[2] Hernando, A., et al. “Origin of Orbital Ferromagnetism and Giant Magnetic Anisotropy at the Nanoscale,” Phys. Rev. Lett. 96 (2006) 057206. Brust, Mathias, et al. “Synthesis of Thiol-derivatised Gold Nanoparticles in a Two-phase Liquid-Liquid System.” J. Chem. Soc., Chem. Commun., (1994) 801.

[3] Yamamoto, Y.; Hori, H., “Direct Observation of the Ferromagnetic Spin Polarization in Gold Nanoparticles: A Review,” Re. Adv. Mater. Sci. 12 (2006) 23-32.

[4] Sorensen, C. M.; Klabunde, Kenneth J., “Magnetism,” Nanoscale Materials in Chemistry (2001) 169-221.

[5] CRC Handbook , 63rd Edition (1982-83) 4-134-4-139. [6] Saunders, A. E.; Sigman, M. B; Korgel, B. A. “Growth Kinetics

and Metastability of Monodisperse Tetraoctylammonium Bromide-Capped Gold Nanocrystals,” J. Phys. Chem. B, 108 (2004) 193-199.

Figure 3: SQUID data ~ 1.8 ± 0.26 nm Au NPs on quartz tubing.

Figure 4: SQUID data ~ 5.5 ± 0.62 nm Au NPs on quartz tubing.


MAT

ERIA

LS

Abstract:Palladium nanoparticles were prepared using colloidal synthesis routes, and their stability was characterized with and without a surfactant. The catalytic activity of these nanoparticles was also determined after various treatments. The palladium colloidal nanoparticles were synthesized through a novel mesityl route, combining palladium (II) chloride with magnesium bromide mesityl to form palladium mesityl and magnesium bromochloride as a precipitate. The palladium nanoparticles formed were stabilized by the surfactant trioctylphosphine (TOP).

After being supported on alumina substrates, the colloidal nanoparticle samples were analyzed for catalytic activity for CO oxidation. Particle sizes were analyzed via TEM, STEM, light scattering and XRD. Samples were examined before and after thermogravimetric analysis (TGA), as well as before/after CO oxidation. The particle size distribution of the Pd/Al2O3 catalyst after reactivity measurements was very similar to that before removal of the surfactant (TOP) indicating that these nanoparticles are stable during CO oxidation. We conclude that the colloidal synthesis method allows for fine control over the Pd nanoparticles size and represents a novel approach to synthesize nanoparticle catalysts.

The Stability and Catalytic Reactivity of Colloidal Palladium Nanoparticles on Al2O3 Supports

David LavensonChemical Engineering, Lehigh University

NNIN REU Site: Nanoscience @ UNM, University of New MexicoNNIN REU Principal Investigator: Prof. Abhaya Datye, Dept. of Chemical Engineering, University of New MexicoNNIN REU Mentor: Ayman Karim, Dept. of Chemical Engineering, University of New MexicoContact: [email protected], [email protected]

Introduction:

Palladium nanoparticles are used in a wide variety of applications such as selective hydrogenation. In this project, our goal was to use these for the methanol steam reforming for H2 production. By synthesizing the particles via colloidal synthesis, we can control the size as well as composition. The colloidal nanoparticles are covered with a surfactant to prevent agglomeration, but the surfactant must be removed prior to reaction. Hence, it is important to investigate the stability of these nanoparticles after exposure to reaction environments.


Synthesis: In these studies, we studied a novel synthesis route using organometallic precursors (mesityl compounds) to synthesize colloidal nanoparticles as shown in the reaction below:

PdCl2 + MgBrMes (magnesium bromide mesityl) ‡ Pd(Mes)2 + 2 MgBrCl

The synthesis was performed under argon atmosphere at all times under a Schlenk line. The sample was refluxed at 300°C for 30 minutes before being exposed to air. The sample was washed with ethanol, centrifuged, and the supernatant was discarded. The nanoparticle suspension in hexane was also deposited on alumina supports via impregnation.

Transmission Electron Microscopy: Scanning and high resolution transmission electron microscopy (STEM and HRTEM) was performed on a JEOL 2010F FASTEM field emission gun STEM. Images were analyzed using Digital Micrograph™ to produce a size distribution. Light scattering was also utilized to produce an initial size distribution before TEM/STEM imaging.

Thermogravimetric Analysis (TGA) - Surfactant Characterization: Thermogravimetric analysis was performed using a TA instrument. These studies used ramp setting of 5°C/min from 25°C to 700°C or 900°C. Most samples ranged between 30.0-40.0 mg in size unless otherwise noted. Samples were run under either air or nitrogen gas at 100 mL/min. Samples were run as Figure 1: STEM of Pd nanoparticles suspended in solution.


MATER

IALS

prepared and after CO oxidation.CO Oxidation: CO oxidation was performed to

investigate the catalytic potential of the sample when it was fresh, as well as after we ran a sample through the TGA. CO oxidation was completed under 1.0% CO and 0.5% O2 in balance helium. Samples of Pd/Al2O3 between 16.0-20.0 mg were used for these studies. A gas chromatograph was used to monitor CO/O2 injections as well as the outlet concentrations of CO/O2/CO2. In accordance with literature values, the sample was run between 150°C and 200°C, the range at which Pd is known to become catalytically active. Data was collected in terms of CO2 conversion. Samples were run as-prepared and after calcination at 700°C at a ramp rate consistent with the TGA experimentation noted earlier.

Results:

TEM and STEM: TEM and STEM imaging for different samples showed very similar particle size distributions. In Figure 1, we can see a wide distribution of particles, all with particles less than 4 nm. Size distributions were collected for the as-prepared sample, the sample after CO oxidation, and after calcination and CO oxidation. The as-prepared sample showed an average of 2.4 ± 0.4 nm, the CO oxidized sample showed 2.6 ± 0.7 nm and the calcination with CO oxidation showed 3.0 ± 1.0 nm.

TGA: TGA of the as-prepared Pd/Al2O3 sample showed the loss of surfactant (TOP) at around 500°C in both air and nitrogen environments (Figure 2). This data was then used for determining a calcination temperature of 700°C to use for preparing samples for CO oxidation after removal of TOP. TGA was also conducted on a sample after undergoing CO oxidation. These results showed there was still trace surfactant remaining that was again burned off by 500°C, but much less remaining compared to the original sample.

CO Oxidation: Figure 3 shows a comparison of two

18.4 mg samples, the as-prepared sample and one after calcination at 700°C in air. The as-prepared sample becomes visibly active on the graph at 190°C and reaches a maximum conversion of about 25%. However, the sample that was calcined shows a dramatic change in conversion after 180°C, reaching 100% conversion at 190°C and holding at 100% till 200°C.

Conclusions:

The surfactant stabilized colloidal palladium nanoparticles were active in their as-prepared state but were found to improve after removal of the surfactant trioctylphosphine, as well as retaining a small particle size throughout the process with no apparent signs of agglomeration. The particle sizes were all less than 3.0 nm, an ideal particle diameter size for nanoparticle catalysts. The TGA results indicated a complete loss of surfactant after 500°C.

Acknowledgements:

I would like to thank Professor Abhaya Datye, Ayman Karim and the CMEM Catalysis lab group for such an amazing summer. Also, the NSF and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Programfor funding, as well as all of my fellow REU students at UNM.

Figure 2, left: TGA curve in air to 700°C.

Figure 3, right:CO oxidation of Pd on Al2O3.

Figure 4: TEM image of Pd nanoparticles.


MAT

ERIA

LS

Introduction:

Dilute nitride (In)GaAsN alloys are useful for applications in infrared laser diodes, high-efficiency solar cells, and high-performance heterojunction bipolar transistors. The addition of N to GaAs-based semiconductors decreases the bandgap energy without drastically affecting the lattice parameter. To date, literature reports have presented substantially lower electron mobilities for dilute nitride semiconductor alloys in comparison with those of (In)GaAs. Furthermore, for (In)GaAsN alloys, the electron mobility has been reported to decrease as the N incorporation increases. At present, the precise role of N in reducing the electron mobility is not well understood. In order to study nitrogen-related electron scattering effects in GaAsN, with minimal contributions from ionized impurity scattering, we are examining the transport properties of modulation-doped AlGaAs/GaAs(N) heterostructures, with Si dopants in the AlGaAs barrier layer spatially separated from the nominally undoped GaAs(N) channel layer. We will discuss the dependence of the mobility on carrier density, with a focus on the insulating phenomena associated with multiple scattering effects past a critical carrier density.


The heterostructures consisted of modulation-doped AlGaAs/GaAs(N) heterostructures grown via molecular beam epitaxy (MBE), on GaAs (001) substrates. An initial 500 nm thick GaAs buffer layer was grown at 580°C. After buffer layer growth, a 50 nm thick GaAs(N) channel was grown at 400°C. Next, a 5 minute pause was used to ramp the substrate temperature to 580°C, and layers of 1 nm GaAs, 20 nm undoped Al0.3Ga0.7As, 60 nm Si-doped Al0.3Ga0.7As, and 10 nm GaAs were then grown in succession. Carriers from the Si-doped AlGaAs layer migrated into the GaAs(N) channel and were confined in a triangular well, producing a two dimensional electron

Effects of N Incorporation on the Electronic Properties of GaAsN-Based Modulation-Doped Heterostructures

Niall M. ManganPhysics and Mathematics, Clarkson University

NNIN REU Site: Michigan Nanofabrication Facility, University of MichiganNNIN REU Principal Investigator: Rachel S. Goldman, Materials Science and Engineering /

Electrical Engineering and Computer Science, University of MichiganNNIN REU Mentors: Matthew Reason, Materials Science and Engineering; Yu Jin, Physics;

Cagliyan Kurdak, Physics; University of MichiganContact: [email protected], [email protected]

gas (2DEG). The undoped AlGaAs layer enabled spatial-separation of the 2DEG and the ionized impurities, thereby minimizing the Coulombic long-range scattering effects on the carriers [2]. The density of free carriers in the channel was further controlled by front gating, or by illumination at low temperature.

Electron transport measurements were implemented with eight-arm gated-Hall bars (1050 x 150 µm), fabricated by standard contact photolithographic processes, with e-beam evaporated Ni/Ge/Au/Ti/Au (200/325/650/200/2000 Å) contacts and Ti/Au (100/ 1000 Å) gates. The mesa and contacts/gate were defined using positive and negative photolithography and a phosphoric acid etch [6]. Prior to gate deposition, the contacts were annealed at 410ºC for 2 minutes in argon gas.

Figure 1: Magnetic field dependence of rxy and rxx, at T = 4.2K for a gate-controlled 2DEG.


MATER

IALS

Magnetoresistance measurements were performed at 4.2K, with the magnetic field swept from 0 to 7 Tesla in a superconducting NbTi magnet. To modulate the free carrier density in the 2DEG, the gate voltage was swept from -60 mV to 150 mV, and data were collected every 10 mV. A near-infrared light emitting diode was also used to illuminate the sample surface, to increase the free carrier density in the 2DEG, through the persistent photoconductivity effect [3].


In Figure 1(a), the minima of the Shubnikov-de Haas oscillations in the magnetoresistance data correspond to the Quantum Hall plateaus in Figure 1(b). Both of these quantum phenomena result from the increasing magnetic field altering the spacing between the Landau levels, thus sweeping the Landau levels with respect to the Fermi level. As the gate voltage decreases from 150 mV to -30 mV, the carrier density (resistivity) decreases (increases), and the Shubnikov de Haas oscillations and quantum Hall plateaus become less apparent, as shown in Figure 1.

From the gated resistivity and Hall effect measurements, we calculated the carrier density and carrier mobility at each gate voltage, shown in Figure 2. One characteristic of the relation between carrier density and mobility was the metal-insulator transition behavior due to multiple scattering effects at carrier densities below a critical carrier density (Nc). Based on the structure and doping level of our 2DEG, we predicted our critical carrier density to be around 7 x 1010cm-2 [4]. The gated 2DEG

data exhibited a deviation from power law dependence, recognized as the critical carrier density, at about 9 x 1010cm-2 as shown in Figure 2.

The relationship between carrier density and carrier mobility for densities above 9 x 1010cm-2 reveals information about the mechanisms of electron scattering by N atoms. The dependence of mobility on carrier density can be expressed as µ~nα [5]. The α values correspond to the slope of a linear-least squares fit to log (µ) vs. log (n). In the control 2DEG, α~1, suggesting remote ionized impurity scattering due to Coulombic interactions between the free carriers and the ionized impurities, is the dominant scattering mechanism. The mobility increases rapidly with n, due to increased screening of the ionized impurity potential. For the 0.08%N 2DEG, α~0.1, indicating ionized long-range scattering is likely not the dominant mechanism. The increased carrier density does not have the same screening effect on neutral scatterers, and therefore does not increase the mobility. For n > Nc, µ is independent of n, similar to calculations which assume N acts as a neutral independent local scatterer [1], as shown by the line in Figure 2. Thus, we tentatively conclude that N acts as a neutral, short-range scatterer.

Acknowledgements:

We gratefully acknowledge the National Science Foundation through grant #DMR-0606406 and the NNIN Grant ECS-0335765, as well as the Goldman and Kurdak Groups and the MNF at the University of Michigan.

References:[1] S. Fahy, A. Lindsay, H. Ouerdane and E.P. O’Reilly, (accepted

to Phys. Rev. B) (2005).[2] S. Hiyamizu, J. Saito, K. Nanbu and T. Ishikawa, Jpn. J. Appl.

Phys. 22, L609 (1983).[3] H.L. Stormer, A.C. Gossard, W. Wiegmann and K. Baldwin,

Appl. Phys. Lett. 39, 912 (1981).[4] A. Gold, Phys. Rev. B 44, 8818 (1991).[5] K. Hirakawa and H. Sakaki, Phys. Rev. B 33, 8291 (1986).[6] Etch mesa with H3PO4:H2O2:H2O at 1:1:25 ratio for 2.5 mins.

Figure 2: Electron mobility as a function of carrier density, measured at 4.2K.


MAT

ERIA

LS

Abstract:The electron transfer reaction between thiosulfate and hexacyanoferrate (III) catalyzed by gold nanostructures has been studied as a function of nanoparticle shape. Gold nanoprisms were prepared on substrates and their catalytic activities were compared to nanoprisms that had been annealed by various methods and to various degrees. The activation energies were also found for both roughened and smooth thin gold films. The activation energies for the reactions were found in order to determine the catalytic ability over the temperature range 23-60°C. An analysis of the results suggests that unannealed surface atoms with unsaturated valency (e.g. those present on unannealed prisms bound to a substrate or on roughened thin film surfaces) have the highest catalytic activities, and annealing of the prisms leads to an increase in the activation energy and thus a lowering of their catalytic efficiencies.

Introduction:

Recent excitement in nanoparticle catalysis has partly been due to the discovery of shape and size dependent nanoparticle properties. A variation in catalytic activity has been reported between nanoparticles of different crystal surfaces and thus different ratios of available atoms to react with on the crystals’ faces, edges, and corners [5]. Colloidally synthesized platinum nanoparticles have shown varying degrees of catalytic activity during the electron transfer reaction between thiosulfate and hexacyanoferrate (III) depending on the nanoparticle’s shape. The activation energies for the reaction increase as the particle shape changes from a tetrahedral, to a cube, and then to a sphere, which correlate with a decrease in catalytic activity [4]. Surface plasmon resonance absorption occurs when incident light oscillates the conduction band electrons of the metal particle and is highly dependent on the particle’s shape, size, surface quality, and medium. For this reason, changes in the particle’s surface plasmon resonance absorption were spectrally followed to qualitatively track

The Effect of Annealing Metallic Nanoparticles on their Catalytic Efficiency

Katrina M. MurphyBioengineering, Oregon State University

NNIN REU Site: Microelectronics Research Center, Georgia Institute of TechnologyNNIN REU Principal Investigator: Mostafa A. El-Sayed, Chemistry & Biochemistry Department, Georgia Institute of TechnologyNNIN REU Mentor: Christopher E. Tabor, Chemistry & Biochemistry Department, Georgia Institute of TechnologyContact: [email protected], [email protected]

the physical changes that occurred during the annealing processes [2].

For this work, nanosphere lithography [1,3] was utilized to make gold nanoprisms stabilized on quartz and annealed to varying degrees in order to study the effect of the nanoparticle’s physical properties in the reaction between hexacyanoferrate (III) [Fe(CN)6

3-] and thiosulfate [S2O3

2-] (shown in Figure 1). All nanoparticles were used uncapped and all reactions were done in triplicate to catalyze the reaction at a 1:15 ratio of [Fe(CN)6

3-]: [S2O32-] at three different temperatures.

Figure 1: The gentle electron transfer reaction between thiosulfate and hexacyanoferrate III.


Gold nanoparticles were initially fabricated using nano-sphere lithography which has been developed within the literature [1,3]. Briefly, quartz slides were cleaned and rendered hydrophilic. A mask was created on the substrate by self-assembling a monolayer of sub-micron polystyrene spheres. Using a PVD filament evaporator, 35 nm of gold was deposited normal to the substrate surface, creating gold nanoprisms within the interstitial areas between spheres. The polystyrene spheres were then removed via sonication in ethanol. By thermally heating some of these nanoprisms at 930°C for 30 seconds in an AET RTP, the particles melted into near spherical nanoparticles. A third sample of nanoparticle catalysts was created by transferring the substrate-bound nanoprisms into DDI water via a femtosecond laser pulse. The nanoparticles were analyzed before and after transfer using SEM and TEM techniques (Figures 2 and 3, respectively) and a slight annealing of the particle’s surface were observed. Thin film gold surfaces were prepared by depositing gold onto clean quartz and smooth mica surfaces for the rough and smooth films, respectively. The surfaces were analyzed by AFM to verify a distinction between film roughnesses.


MATER

IALS

The rate of the reaction was measured by spectrally following the decrease in the concentration of the reactant hexacyanoferrate (III) at 420 nm. The rate was assumed to be pseudo-first order by using a 15:1 ratio of [S2O3

2-]: [Fe(CN)6

3-], and rate constants were obtained using the Beer-Lambert Law for 23, 45, and 60°C. The natural logs of the rate constants were plotted against inverse temperatures to produce an Arrhenius plot (Figure 4) from which the activation energies for each gold catalyst were obtained.


By analyzing different surfaces of nanoparticles and thin films in the chemical reaction in Figure 1, we have shown that surfaces exposed to the least amount of annealing have large catalytic activities. The substrate bound prisms were the most catalytically active with an activation energy of 30.2 ± 2.6; followed by prisms suspended in solution, 33.0 ± 3.4; rough thin film surface,

37.1 ± 2.5; spheres supported on the quartz, 37.9 ± 3.2; the smooth thin film surface, 48.4 ± 1.9; and lastly the uncatalyzed reaction, 56.7 ± 5.0. Consequently, the nanoparticles with rougher surfaces and thus unsaturated atoms were better catalysts than surfaces that were annealed by heating, which results in saturated atomic valency.

Future Work:

Additional investigation for this research includes examining the recycling potential of these gold nanoparticles, using same-sized and same-shaped nanocatalysts in other types of chemical reactions, and continuing to control the shape and size of the nanoparticles while making them smaller.

Acknowledgements:

Sincerest thanks to Chris Tabor, Dr. Mostafa El-Sayed, the entire Laser Dynamic Laboratory, Jennifer Tatham, Melanie-Claire Mallison, the patient cleanroom staff, the GSI group, and the MiRC. The Natioanal Science Foundation provided funding for this research through the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] El-Sayed, M. A. Chem. Soc. Rev., 2006, 35, 209. [2] Schatz, G. J. Phys. Chem. B 2003, 107, 668. [3] Van Duyne, R. J. Phys. Chem. B 2001, 105, 5599. [4] El-Sayed, M. A. J. Phys. Chem. B 2005, 109, 12663.[5] Somorjai, G. A. J. Catal., 1987, 103, 213.

Figure 2: SEM images of a) nanoprisms on quartz, and b) nanospheres on quartz.

Figure 4: Arrhenius plots were used to derive the activation energy.

Figure 3: A TEM image of nanoprisms in deionized water.


MAT

ERIA

LS

Abstract:Zinc oxide nanowires grown from aqueous solutions of methenamine and zinc nitrate are used in dye-sensitized solar cells. The growth mechanism of the ZnO nanowires and the solution chemistry is not yet well understood. During the reaction, heterogeneous growth of nanowires is limited by homogenous nucleation of ZnO particles. The reaction kinetics were studied by complexometric titration. It was determined that homogeneous nucleation of ZnO particles is first order in methenamine and zinc nitrate concentration. The kinetics of the heterogeneous growth of ZnO nanowires was also studied to determine if the nanowire growth follows the same kinetics as the homogenous ZnO particle growth.

Introduction:

Dye-sensitized solar cells (DSSCs) have the potential to convert solar power to electricity efficiently and at a low cost, making it promising solar cell architecture for large scale solar energy implementation. There are three main components to a DSSC; (1) a ~10 µm thick film of wide band gap semiconductor nanoparticles such as TiO2 or ZnO nanoparticles, (2) a monolayer of organic dye molecules absorbed onto the semiconductor, and (3) a liquid electrolyte containing the redox couple I-/I3

- which interpenetrates the dye-coated nanoparticles. When a photon excites the electron in the dye, it is injected into the semiconductor and is carried to the anode and through the load to the cathode where it reduces the I- in the electrolyte. The I3

- then regenerates the dye thus completing the circuit.

In conventional nanoparticle DSSCs, the electrons diffuse to the anode by hopping 103-106 times between particles [1]. With each hop, the electron can recombine with the electrolyte. The diffusion rate and recombination rate are both on the order of milliseconds, allowing recombination to limit the efficiency of the DSSC. However, ZnO nanowire DSSCs provide a direct path

Solution-Growth of Zinc Oxide Nanowires for Dye-Sensitized Solar Cells

Yasuhide NakamuraChemistry and Chemical Engineering, University of California Santa Barbara

NNIN REU Site: Minnesota Nanotechnology Cluster, University of Minnesota, Twin CitiesNNIN REU Principal Investigator: Dr. Eray Aydil, Chemical Engineering, University of Minnesota, Twin CitiesNNIN REU Mentor: Janice Boercker, Chemical Engineering, University of Minnesota, Twin CitiesContact: [email protected], [email protected]

to the anode, which increases the diffusion rate without increasing the recombination rate and could therefore increase the efficiency of DSSCs.

In the solution growth of ZnO nanowires, two types of crystal nucleation compete. Homogenous nucleation produces undesired ZnO particles and heterogeneous nucleation produces nanowires. In the existing procedure, the homogenous nucleation of ZnO particles dominates and depletes the precursor limiting the wire growth. To understand the kinetics and to hypothesize a mechanism that replicates the data, the rate of Zn2+ depletion was observed with time and various initial concentrations of reagents.


To determine the Zn2+ concentration, complexometric titration with ethylenediaminetetraacetic acid (EDTA) was used [2]. First, homogenous nucleation of ZnO was monitored. The reaction took place in 15 ml centrifuge tubes with 10 ml of DI water containing 0.016 M of Zn(NO3)2 and 0.025 M of methenamine. The tubes

Figure 1: SEM image of ZnO nanowires at 240 minutes.


MATER

IALS

were placed in an oven set at 95°C and were removed at various time intervals. The tubes were centrifuged at 3000 rpm for 30 minutes to remove suspended ZnO particles from the solution, which interfere with the titration. Then 1 ml of the solution was placed in a beaker with 25 ml of DI water and 5 mL of pH 10 ammonium buffer. While stirring vigorously, a drop of eriochrome black T indicator was added and the solution was titrated with 0.002 M EDTA. The procedure was repeated with 0.050 M and 0.0125 M methenamine to see the effect of methenamine concentrations.

It was hypothesized that the depleted Zn2+ became ZnO particles and there was no unaccounted for Zn2+. To justify this statement, particles were collected from the tubes and weighed. The structure of particles was also found using x-ray diffraction.

ZnO nanoparticles were synthesized by mixing of 0.001 M of zinc acetate and 0.0016 M of NaOH at 35°C with isopropanol as solvent [3]. The nanoparticles were rocker coated onto (100) silicon substrates. The substrates were annealed at 450°C for 30 minutes. Then, substrates were washed with DI water. The substrates were coated, annealed, and washed again.

Heterogeneous nucleation was monitored by placing the coated silicon substrates inside the centrifuge tubes. When the tube was removed from the oven, the substrates were taken out and washed. Horizontal views of the substrates were taken with SEM JEOL 6500 model, see Figure 1. The lengths of nanowires were measured and average nanowire lengths were determined as a function of time.


Figure 2 shows no significant change in Zn2+ concentration was observed until 50 minutes. The temperature profile of the solution indicated that the

solution comes to its final temperature of 95°C at this time. To remove the temperature effect, rate constants were calculated by finding the slope of the ln[Zn2+](T)/ [Zn2+](T0) versus (T-T0), where T0 is the time 95°C was reached. The contribution of Zn2+ concentration to the depletion rate was determined to be first order. Methenamine concentration was also determined to be first order. As seen in Figure 3, the rate constants change with initial concentration of methenamine.

When the heterogeneous reaction was done simultaneously in the tube, similar rate constants were derived. It is safe to conclude that heterogeneous did not interfere with the homogeneous reaction. As can be seen in Figure 4, the average wire length increases with time, but its kinetic behavior cannot be concluded.

Future Work:

The rest of the substrates that were made must be imaged with SEM to make more graphs similar to Figure 4, which shows the wire growth over time. The data will then be compared with the titration data to discuss the possible relationship between heterogeneous and homogenous nucleation.

Acknowledgments:

The author would like to thank Dr. Eray Aydil, Janice Boercker, the NSF and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Baxter, J. B.; Walker, A.; Ommering, K.; Aydil, E. Nanotechnology

June 2006, 17.[2] Boercker J. Synthesis of ZnO Nanowires and TiO2 Nanowires

for DSSC, April 2006.[3] Searson et. al. J. Coll, Inter. Science 2003,


MAT

ERIA

LS

Introduction:

The current design of P- and N- doped silicon solar cells lacks the efficiency to adequately contribute to the existing energy demand. Because of this, researchers are exploring new solar cell designs. The design explored in this paper is called “Nanostructured Inorganic Donor/Acceptor Photovoltaics” and is intended to be an improvement on similar work recently published in Science [1].

In this design, an electron current is produced by the donor/acceptor photovoltaic method of electron diffusion between CdTe and CdSe. These semiconducting materials were selected because the CdTe has conduction and valance bands that are offset in energy compared to those of CdSe in such a way that electron and hole diffusion are optimized but recombination is minimized. Solar radiation excites electrons to the conduction band but they aren’t free to move because they are attracted to the hole they left behind. The electrons in the CdTe conduction band are freed by falling to the lower energy conduction band of CdSe. The electron does work in a circuit and is returned to it’s hole to start the process over again.

The design of the device presented here consists of a large array of CdSe rods infiltrated with CdTe quantum dots, as schematically illustrated in Figure 1. This report focuses on the fabrication and characterization of CdSe nanorods. The design provides a very high surface contact area for the CdSe nanorods and the CdTe quantum

Cadmium Selenide Semiconducting Nanorods Vertically Aligned to a Conductive Substrate for Solar Cell Application

Emily NorvellMaterials Engineering, California Polytechnic State University, San Luis Obispo

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigators: Martin Moskovits, Galen Stucky, Department of Chemistry

and Biochemistry, University of California Santa BarbaraNNIN REU Mentor: Martin Schierhorn, Department of Chemistry and Biochemistry, University of California Santa BarbaraContact: [email protected], [email protected]

dots material. Furthermore, the nanoscale of the design boasts a short distance for the electron to travel to be separated from its hole, and the organized structure of the rods allows a direct path for the freed electron to travel to the electrode, so recombination is minimized.

Procedure:

A solution of H2O, H2SO4, 0.3 M CdSO4, 2.8 mM SeO2 was prepared and used to deposit in an electrochemical cell with a Pt mesh as the cathode. The anode was nanoporous alumina oxide (PAO) with a conductive backing which was a template for the formation of rods. We achieved the best results when we evaporated at least 10 µm of Ti before we evaporated the conductive backing of either Ag, Au or Pt and when we had the conductive back on the barrier layer side. We deposited in the PAO template using a method called cyclic voltammetry to achieve a 1:1 Cd:Se ratio so that the rods could be composed of a pure material. If straight DC voltage is used, an excess of Se will be deposited. The voltage was swept between -357 mV to -757 mV vs. a Ag/AgCl reference electrode. CdSe and Se were deposited from -357 mV to -736 mV. At -736 mV, Cd2+ was reduced, depositing an excess of Cd in the template which reacted with the Se precursor to form CdSe. At -656 mV, all the Cd+ that didn’t form CdSe was oxidized, leaving a layer of CdSe. The CdSe was deposited layer by layer for either 600, 1200, or 1800 cycles.

Results:

An approximate 1:1 ratio of Cd:Se was confirmed with energy-dispersive x-ray (EDX) for the 0.3 M CdSO4: 2.8 mM SeO2 ratio (almost 100 fold excess of Cd ions). This ratio means our rods were made of pure CdSe. The length of the rods was controlled by the number of voltage cycles. 1200 cycles yielded ~ 1 µm.

Rods deposited in the alumina templates stood unsupported and did not clump together when the template was etched away with NaOH and dried in super critical CO2. Nanorods that are vertically aligned to the

Figure 1: Side view of our nanostructured inorganic donor/acceptor photovoltaics design.


MATER

IALSconductive substrate can be seen in Figure 2.

In preliminary work, CdSe was deposited onto slides of Pt. When the CdSe films were annealed at 300°, 400°, 500°, and 600°, an increase in characteristic crystalline CdSe peaks was shown with analysis of the slides with x-ray diffraction (XRD). XRD was done on the rods but inconclusive data resulted due to the random orientation of the crystallinity of the rods.

As-deposited rods were examined with a transmission electron microscope (TEM), which can be seen in Figure 3. The rods appear grainy, which was confirmed with selective area electron diffraction (SAED) to mean that the rods are polycrystalline, characterized by rings on the diffraction pattern. The spacing of the rings matched those of hexagonally oriented CdSe. Rods were annealed at 600°C and examined with TEM, shown in

Figure 4. They look more uniform and smoother than the unannealed rods. SAED showed bright points on the diffraction pattern which indicated that annealed rods were more crystalline. These results confirmed that the crystallinity of CdSe nanorods increased when annealed.

Future Work:

As more data is collected, a ratio of length per voltage cycles can be determined. Different materials for semiconductor rods and conductive backings will be tested for their effectiveness. Eventually solar devices can be assembled and tested for their efficiency.

Conclusion:

We successfully produced CdSe rods with the desired characteristics towards the production of photovoltaic devices with potentially greater efficiencies than current solar cells. We were able to produce pure CdSe material from Cd and Se precursors in solution. Vertically aligned rods were fabricated that made good contact to a conductive substrate and provided a large surface area for contact with sensitizing quantum dots. The as-deposited rods were poly crystalline and became more single crystal after annealing, which was desired for good electron transport to the electrode.

Acknowledgements:

Thanks to Martin Moskovits, Galen Stucky, Martin Schierhorn, UCSB, NSF, the ICB and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] I. Gur, N. A. Fromer, M. L. Geier, A. P. Alivisato, Science. 462,

310 (2005).[2] Q. Li, M. A. Brown, J. C. Hemminger, R. M. Penner, Chem.

Mater. 3432, 18 (2006).

Figure 2: Nanorods vertically aligned to conductive substrate.

Figure 3: TEM of unannealed rods (inset): SAED of rods. Figure 4: TEM of annealed rods (inset): SAED of rods.


MAT

ERIA

LS

Abstract:Inorganic nanoparticles have unique nanoscale optical and electrical properties independent of their bulk systems. This research focused on chemical modifications of gold nanoparticle arrays to manipulate conductivity of films of these nanoparticles. Organic ligand chains attached to nanoparticle cores were crosslinked using dithiols to increase conductivity. The temperature dependant conductivity of nanoparticle films as a function of chemical crosslinking was measured, and it was found that crosslinking ligands enhance conductivity, as does increasing temperature.

Introduction:

The gold nanoparticles (AuNPs) used in this research have an approximate diameter of 2 nm: this particle size determines the single electron charging energies in nanoparticle films and governs the AuNP’s electrical properties. It is important to be able to understand, characterize, and model the behavior of gold nanoparticles, especially the electronic properties as a function of chemical modification. With this knowledge, it would be possible to manipulate AuNPs to have well defined, stable properties for use in electronic or semiconductor devices [1].


In this project, AuNPs were chemically modified in an effort to manipulate the conductivity of their corresponding films as spin-coated on top of electrodes. The nanoparticles were solution synthesized using the Brust method and the gold cores were stabilized with insulating hexane thiolate (C6) ligands [2]. Nanoparticles were kept in concentrated toluene or heptane solutions (~350 mg NPs/mL solvent). The conductivity of the nanoparticles was measured by making thin AuNP films on electrodes and measuring the resulting current as a voltage was swept across the metal contacts (-4 V to 4 V). Nanoparticle solutions were spin-coated (2000 rpm)

Chemical Crosslinking and Temperature Dependant Conductivity of Ligand-Stabilized Gold Nanoparticles

Stephanie PetrinaMaterials Science and Engineering, Virginia Polytechnic Institute and State University

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigator: Dr. Galen D. Stucky, Chemistry and Biochemistry, University of California Santa BarbaraNNIN REU Mentor: Shannon Boettcher, Department of Chemistry and Biochemistry, University of California Santa BarbaraContact: [email protected], [email protected]

into thin films (~300 nm thick) on small electrical devices such as interdigitated electrodes (IDEs) or planar electrodes made in lab. For the planar “sandwich” electrodes, a thin gold film was evaporated on top of the AuNP film to form a top contact. The NP films were crosslinked using dithiols to interlink the AuNP cores and the conductivities of the films were measured at temperatures from 80 to 320 K using a liquid nitrogen vacuum cryostat to control the temperature.

Crosslinking:

The goals of this research were to develop a mathematical model to describe the electrical effects of crosslinking and to understand the mechanisms of charge transport in AuNP films. To crosslink AuNPs via bifunctional organic ligands, the nanoparticle films were submerged in dithiol solutions from 30 minutes up to four hours. Over time, dithiols in solution replaced more of the hexane thiols attached to AuNP cores. Nonanedithiol (NDT) and benzenedithiol (BDT) were used to crosslink ligands in concentrations of 5 µL NDT per mL isopropyl alcohol and 5 µg BDT per mL toluene.


For IDEs, current-voltage plots always displayed ohmic behavior. Current was a constant function of bias, and conductivity only increased with increasing temperature. Nanoparticle films on sandwich electrodes did not display ohmic behavior after biases much larger than ± 0.2 V were applied, as can be seen in Figure 1. Investigation of this behavior is necessary to understand what mechanism causes this drastic increase in conductivity.

The amount of crosslinking, and thus conductivity, was directly related to the amount of time spent in solution. Data characterizing the temperature dependant nature of this relationship could be fit to an Arrhenius equation where EA is activation energy, R is the gas constant


MATER

IALS

(8.3144 J/mol K), and T is the temperature in Kelvin. Low field conductivity data, where the current-voltage relationship remained linear, was fit to this model as can be seen in Figure 2.

Because distinctly different current-voltage relationships were observed at low biases versus those at high biases, the temperature dependant data was divided into low (-0.2 V to 0.2 V) and high (4 V) field conductivity models for analysis. In Figure 2, low field conductivities of the film as uncrosslinked, crosslinked in NDT solution, and crosslinked in BDT solution, were plotted together. The activation energy decreases slightly from the uncrosslinked film (EA = 15.4 kJ/mol) to NDT crosslinking (EA = 14.7 kJ/mol), but the significant decrease in activation energy is observed after BDT crosslinking (EA = 10.5 kJ/mol). The double bonds in the benzene ring in BDT allowed for significantly faster electron transfer, and thus, much higher conductivities were observed. According to the Arrhenius equation, electron hopping is a logarithmic function of temperature, so the conductivity shows extremely strong temperature dependence. The activation energy for an electron to “hop” to an adjacent nanoparticle provided a quantitative parameter to compare the effects of crosslinking.

In Figure 3 it is evident that high field conductivities deviate from the Arrhenius model. To model the high field conductivity behavior, a variable range hopping (VRH) model was used, where electrons were theoretically able to jump more than one AuNP core or “potential well” at a time. As a result, conductivity became a linear function of the square root of inverse temperature, as opposed to only inverse temperature, as in the Arrhenius model.

Future Work:

A comprehensive model for the low and high field conductivity must be verified. The ultimate goal of this nanoparticle characterization is to be able to understand the charge transport mechanisms in order to control electrical AuNP properties. This knowledge could then be applied to solution synthesized semiconductor nanoparticle systems with optoelectronic and photovoltaic applications.

Acknowledgements:

Thanks to my mentor, Shannon Boettcher, for his continuous guidance and support. I appreciate Dr. Galen Stucky and the Stucky group graciously welcoming me into their lab, as well as the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for funding this research.

References:[1] (a) Joseph, Y.; Besnard, E.; Rosenberger, M.; Vossmeyer, T. J.

Phys. Chem. B 2003, 107, 7406-7413. (b) Templeton, A.C.; Hostetler, M.J.; Kraft, C.T.; Murray, R.W. J. Am. Chem. Soc. 1998, 120, 1906-1911. (c) Templeton, A.C.; Wuefling, W.P.; Murray, R.W. Acc. Chem. Res. 2000, 33, 27-36. (d) Wuefling, W.P.; Green, S.J.; Pietron, J.J.; Murray, R.W. J. Am. Chem. Soc. 2000, 122, 11465-11472.

[2] Schaaf, T. G.; Shafigullin, M.N.; Khoury, J.T.; Whetten, R.T. J. Phys. Chem. B 2001, 105, 8785-8796.

Figure 1: Conductivity of uncrosslinked AuNP film at low temperatures.

Figure 2: Arrhenius analysis of low field conductivity on a sandwich electrode.

Figure 3: Low and high field conductivity of an uncrosslinked film.


MAT

ERIA

LS

Abstract:This study modeled epitaxial crystal growth of atomic structures using PMMA colloids with depletant polymer. The suspension was sedimented onto square lattice templates to simulate substrate surfaces. Different types of substrates were modeled by using templates of varying lattice constants. Analysis showed that the templates force square packing instead of hexagonal close pack (hcp). However, the effect can be diminished if the lattice constant is larger than the natural packing distance of the colloids. Square templates with Gaussian noise were also fabricated and preliminary results have shown forced square packing, but not at the same magnitude as non-noisy lattices.

Introduction:

When performing epitaxial growth of atoms, it is advantageous to control the morphology of film growth as much as possible. While it is difficult to perform experimentation upon actual atoms due to their size and high kinetic energy, atomic behavior can be mimicked using colloidal particles. Their ability to self-assemble and, in the presence of depletant polymer, crystallize in an atomic matter makes colloids an ideal candidate for experimentation. This study capitalized upon colloids’ ability to exhibit such behavior.

Experiment:

Micropatterned templates were fabricated using plasma-cleaned coverslides that were coated with PMMA, anti-reflective coating (ARC), and photoresist (PR). Using photolithography methods, arrays of 1 µm circles with three different lattice constants (l.c.) (1.212µm, 1.362 µm, 1.412 µm) were projected onto the PR of the glass slide. The PR was then developed away, exposing regions of ARC and PMMA, which were then etched off using oxygen plasma. Finally, the PR was removed, leaving a pattern of holes in a layer of pure PMMA. An identical procedure was carried out upon another set of coverslides that were instead patterned with a square lattice that contained program-generated

Epitaxial Crystal Growth of Colloids with Short Range Attraction

Erica PrattMechanical Engineering, Biomedical Engineering, Carnegie Mellon University

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Dr. Itai Cohen, Physics, Cornell UniversityNNIN REU Mentor: Sharon Gerbode, Physics, Cornell UniversityContact: [email protected], [email protected]

random Gaussian noise. PMMA particles averaging 0.9 µm in diameter were

stained using a rhodamine solution. These particles were then experimentally density mismatched until the particles experienced Brownian motion while in freefall. Polystyerene polymer was also added to the solution. When the PMMA particles were closer than the radius of gyration of the polymer, the osmotic pressure around the two colloids becomes unbalanced, causing a net attractive force between them. This phenomenon is described by:

where Udep is the depletant potential, σ is the diameter of the particle, and IIp is the osmotic pressure of the polymer. Voverlap is the volume of the overlapping depletion zones between particles at a separation. Confocal microscopy was then used to observe the dynamically changing system while the PMMA particles sedimented onto the templates.

The sedimentation container consisted of a coverslide with a micropipette tube glued to the slide using a combination of UV glue and epoxy. The density mismatched particles were then inserted. Samples were analyzed using confocal microscopy two days after dispersion of colloids into the container. Using confocal

Figure 1: [L-R] Colloids without depletant exhibiting hexagonal close packing (hcp); Island-like growth of oxygen and nitrogen in epitaxial growth; Island-like growth of colloids after depletant effect.


MATER

IALS

microscopy, the samples were scanned both in the horizontal and vertical direction to locate sites of crystal growth. After these sites were located, particle tracking was performed and the pair correlation function, g(r), was calculated. G(r) is the normalized probability of finding a particle a radial distance away from another particle. The form of the g(r) contains information describing the structure of the ordered phase.

Results & Discussion:

Sedimentation upon 1.412 µm Lattice Constant Template: G(r)s for colloidal sedimentation upon a flat surface showed that packing was hcp while colloidal sedimentation upon the 1.412 µm template was square. This order was shown to persist through at least two layers. Further analysis of the pair correlation functions showed that the peaks were consistent with predicted radial values for a square lattice with 1.412 µm spacing. However it was visually observed that the affect of the template diminished as the confocal microscope scanned through z at sites of crystal growth. 1.412 µm lattice spacing is larger than the natural packing distance of the PMMA particles and as layers of particles sediment above the template, the structure began to revert back to hexagonal close-packing.

Sedimentation upon 1.212 µm Lattice Constant Template: The 1.212 µm template also showed that the colloids were conforming to the square lattice. The 1.212 µm template, however, showed no signs of lost ordering when visually observed using confocal microscopy. This would seem to indicate that a spacing of 1.212 µm is

closer to the natural packing distance of PMMA particles than 1.412 µm.

Sedimentation upon Noisy Lattice: The noisy lattice was created by adding random Gaussian noise to the original 1.362 µm square lattice. First layer g(r)s showed similar ordering to non-noisy lattices. Since the noise added was subtle, this was not unexpected. However, in the second layer the subtle stacking disorder of the colloids began to propagate resulting in a g(r) graph that showed that while square packing was evident, it was not as ordered as the standard square lattices. This demonstrates that even a substrate surface with minor variations can have an impact on developing film layers.

Future Work:

While this project has begun to simulate the possible effects of the substrate surface on epitaxial growth, the picture is not yet complete. Further experiments need to be done investigating the effects of the noisy lattices as these more accurately model a realistic substrate. Further work will also be done to see at what level the disorder diminishes in noisy lattices and, as in the case of the 1.412 µm lattice, a lattice with a large lattice constant.

Acknowledgements:

I would like to thank my Principal Investigator Itai Cohen and mentor Sharon Gerbode for being interactive, supportive and knowledgeable advisors. I would also like to thank the Cohen lab and CNF staff along with Jalina Keeling for density mismatching the PMMA particles used in this experiment. Finally, many thanks to the Intel Foundation and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] S. Asakura and F. Oosawa, J. Chem. Phys. 22, 1255-56, 1954.[2] S. M. Inlet A. Oorock, W. C. K. Poon and P. N. Pusey, Physical

Review E, 51:2, 1344, 1995.[3] J. Hoogenboom, D. Derks, P. Vergeer, A. van Blaaderen J. Chem

Phys. 117, 24, 2002.

Figure 2: Colloidal sedimentation: 1) Smooth wall, 2) 1.212 µm l.c., 3) 1.412 µm l.c., 4) Random lattice based on 1.362 µm template.

Figure 3: 1) G(r) for sedimentation on a smooth wall, 2) G(r) for 1.212 µm lattice, 3) G(r) for 1.412 µm lattice, 4) G(r) for noisy template.


MAT

ERIA

LS

Introduction:

Nanoporous gold (NPG) thin films, less than 1 µm in thickness, have been of great interest in large part due to the fact that such materials show great promise for use in diagnostic applications [1], MEMS devices [2], or other applications that require inertness, conductivity, or increased surface area. While previous research has focused on modeling the growth of such films [3,4], there have not been many studies devoted to their mechanical properties. A knowledge of the mechanical properties of NPG films is fundamental to an understanding of issues regarding structural integrity in devices which employ such films.

The focus of this project has been to optimize the processing of NPG films on silicon substrates and to measure the induced stress in the porous gold films by the dealloying method used in the fabrication process. Changes in stress as a function of temperature were also

Fabrication and Characterization of Nanoporous Gold Thin Films

Leila Joy RobersonBiochemistry, Samford University

NNIN REU Site: Cornell NanoScale Science & Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Dr. Christopher (Kit) Umbach, Materials Science and Engineering, Cornell UniversityNNIN REU Mentor: Totka Ouzounova, Materials Science and Engineering, Cornell UniversityContact: [email protected], [email protected]

measured. Scanning electron microscopy (SEM) was used to determine porosity changes due to exposures in various concentrations of nitric acid, and as a function of annealing. Surface-enhanced Raman (SERS) spectra of thiol self-assembled monolayers (SAMs) were acquired using a near-IR confocal Raman microscope with an excitation wavelength of 785 nm.

Fabrication:

NPG thin films (Figure 1) were prepared by selective leaching of silver from an alloy deposit consisting of Ag70Au30 [3] resulting in a mesoporous metal having pore dimensions of 10-30 nm. Three-inch Si wafers were cleaned using a standard metal-oxide semiconductor (MOS) cleaning process for 30 minutes and rinsed with ultrapure (18 MΩ-cm) water. After undergoing an additional rinse/dry cycle, the wafers were placed in a 1000°C thermal oxidation furnace for an hour to facilitate growing of 300 nm of oxide on the substrate. Upon cooling, an adhesion layer consisting of 30 nm Cr / 50 nm Au was evaporated onto the wafer, and approximately 250-300 nm of a Ag70Au30 alloy was deposited using an argon-ion sputter system. Wafers were then immersed in various concentrations of nitric acid (ranging from 15% to 40% v/v), rinsed with DI water, blown dry with nitrogen gas, and allowed to dry overnight. Substrates prepared in this manner were a dark brown in color, with increasing darkness upon immersion in higher concentrations of acid.


Stress Analysis: Stress measurements were made using a common laser scanning technique [5], which measures substrate curvature changes associated with thin film stresses. Initially, it was believed that etching using nitric acid would result in increased stress in the porous film due to the fact that removal of silver atoms from the alloy would create a non-equilibrium level of vacancies in the vicinity of some gold atoms. In order to produce a configuration closer to equilibrium, the gold atoms would aggregate to form clusters [3]. Thus the film

Figure 1: SEM images of nanoporous gold; (a) etched in 15% v/v HNO3, (b) etched in 20% v/v HNO3 (imaged at an angle), and (c) after 9 hours of annealing at 450ºC (imaged at an angle).


MATER

IALS

would become denser and pull on the Si wafer, producing a measurable change in the substrate curvature and results in an overall increase in tensile stress of the gold. However, the opposite was found to occur-the tensile stress decreased after etching. Average stress values measured in the sputtered film were around 340 MPa, which dropped to 60 MPa in the porous film after the etching process. Relaxation of stress also occurred as a result of annealing (Figure 2). Thermal diffusion of gold atoms allows atoms to relieve local stresses through rebonding.

Morphology: Pore and ligament evolution were monitored as a function of annealing at 450°C for 15 hours. Statistical analysis of pore openings using SPIP image processing software showed; (1) that the first 10 hours of annealing resulted in a near-exponential growth in the average pore width, and (2) from 10-15 hours, the average pore width appears to be closing (Figure 3). In addition, the entire process of annealing results in a linear increase in the ligament width. This evolution is also accompanied by ligament coarsening, which affects the material’s ability to be used for surface-enhanced spectroscopy (see Raman discussion).

Raman: SERS is a technique that uses inelastic (Raman) scattering of monochromatic light for the detection of adsorbed molecules on roughened metal surfaces (typically gold or silver). The increase in the Raman scattering rate associated with the SERS measurement can arise either from electromagnetic or chemical enhancement effects, or both [6]. In this experiment, the vibrational modes of 1-dodecanethiol and 2-mercaptopyridine SAMs chemisorbed onto NPG substrates etched in various acid concentrations were analyzed using Raman. These molecules were chosen

because of their large inherent polarizability that results in strong Raman scattering.

Enhancement for both molecules was observed on NPG substrates formed by etching in 15% and 20% v/v HNO3 concentrations, but was not visible with increased acid concentrations (Figure 4). This observation may be due to greater quantities of residual silver in the films etched at lower acid concentrations, since silver is known to be a better SERS substrate. Such enhancement was not visible on substrates that were annealed before thiol deposition. Nanoscopic surface irregularities that are present in many SERS substrates produced by wet-chemical roughening procedures disappear upon annealing.

Acknowledgements:

Many thanks to the Intel Foundation, NSF, and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for funding; Dr. Kit Umbach and Totka Ouzounova; Cornell NanoScale Facility staff; Dr. Brian Gregory, Samford University, for help with the Raman spectra decoding.

References:[1] H. Azzazy, M. Mansour and S.C. Kazmierczak, Clinical

Chemistry, 2006, 52, 1238-1246.[2] M. Cortie and E. van der Lingen, Gold 2003, 28th September-1st

October 2003, Vancouver.[3] J. Erlebacher, M. Aziz, A. Karma, N. Dimitrov and K. Sieradzki,

Nature, 2001, 410, 450-452.[4] J. Erlebacher and K. Sieradzki, Scipta Materialia, 2003, 49,

991-996.[5] W. D. Nix, Mechanical Properties of Thin Films, Met. Trans. A,

1989, 20A, 2230.[6] R.L. McCreery, Raman Spectroscopy for Chemical Analysis,

(Wiley-Interscience, New York, 2000).

Figure 2: Stress-temperature measurement of porous gold film. A relaxation of tensile stress is occurring due to the fact that diffusion of the gold atoms allows local rebonding.

Figure 3: Porosity changes due to annealing at 450ºC for 15 hours.

Figure 4: Raman spectra of 2-Mercaptopyridine. Top: NPG substrates etched in 15% and 20% v/v HNO3 before exposure to 1mM thiol solution. Bottom: substrates etched in 25% v/v HNO3 or greater or subjected to annealing.


MAT

ERIA

LS

Abstract:The alignment of carbon nanotubes using a ferrofluid is explored. By applying a colloidal liquid suspension of ferrous nanoparticles to a wafer with CVD-grown carbon nanotubes, we observe evidence of a mechanical interaction between the nanoparticles and nanotubes facilitated by an externally-applied magnetic field. Future use of this interaction may facilitate the alignment of carbon nanotubes in micro-electronic devices.

Introduction:

Carbon nanotubes (CNTs) have been aligned using several unique methods. For example, Ural’s method [1] uses an electric field to align the CNT’s during CVD growth. However, this method requires electrodes to be placed onto the wafer which take up lots of space where more CNTs could’ve been grown. For multi directional growths, it may be difficult to use this method, because of effects that may stem from the interaction of electric fields within the wafer area. In another method, Liu [2] has prepared chemically-patterned wafers that favor, or oppose, the union of CNTs in liquid suspension. However, one cannot control the length of which CNT’s adhere to the chemically patterned surface.

With this method, we attempt to use ferrofluid (i.e. a colloidal suspension of surfactant-coated, nanometer-sized, ferrous particles, suspended in a carrier liquid) to forcibly align CVD-grown CNTs using a mechanical “raking” of the magnetic nano-particles across the CNTs, as shown in Figure 1.

Ferrofluidic Alignment of Carbon Nanotubes

Cary SmithPhysics, Jackson State University

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell UniversityNNIN REU Principal Investigator: Prof. Sandip Tiwari, Charles N. Mellowes Professor in Engineering,

Electrical & Computer Engineering, Cornell UniversityNNIN REU Mentor: Jay S. VanDelden, Visiting Scientist, Electrical and Computer Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]


Silicon (100) p-type wafers (100 mm diameter, 525 µm thickness, (0.2 Ω-cm resistivity) were MOS-cleaned (NH3O4 & HCl) and spin-rinsed in a Verteq superclean 1600 dryer. Afterwards, ~210 nm of wet-HCl oxide was grown in 28 minutes at 1000°C.

The application of HMDS followed by Shipley S1805 resist at 4000 rpm for 60 sec, resulted in a resist coating of ~700 nm. A soft-bake at 90°C for 60 seconds followed. Immediately after, a 0.29-sec exposure in the GCA AutoStep 200 stepper was done. A post exposure bake at 115°C for 60-sec followed by a 150-sec development in the MIF300 using the Hamatech-Steag opened holes for our catalyst islands. De-scum was completed using an oxygen-clean recipe (50 SCCM O2, 30 mTorr, 150W, 10°C) in the Oxford-80 RIE system.

Using a CVC-4500 e-beam evaporator ~5Å of iron was deposited across the wafers, followed by scribing/breaking into individual chips, and acetone lift-off (10-min) with a final isopropanol rinse.

Our substrates were then inserted into a 1-inch diameter quartz tube furnace (Lindberg/Blue Model #HTF55122A) and ramped to 700°C while flowing 0.80L/min of argon. Purging of the tube occurred for 10-min with 0.80L/min of argon and 0.15L/min of hydrogen at 700°C. Adding 5.5 cm3/min of ethylene (C2H4) for 6-min resulted in carbon nanotube growth. Cool-down at 12°/min with 0.80 L/min flow of argon completed the process. Finally, a droplet of EMG 707 ferrofluid (FerroTec) was applied with a pipette onto the top surface of each chip and a neodymium-iron-boron permanent magnet (~2000G) was slid underneath. The ferrofluid applied to each chip reacted to the magnetic field by flowing controllably across the surface, thereby producing a mechanical “raking” of the CNTs. Following 5-10 swipes of the magnet, the chips were rinsed in a beaker of de-ionized water to remove any remaining ferrofluid.

Figure 1: Alignment of carbon nanotubes using a ferrofluid.


MATER

IALS


SEM analysis of the CNTs using a Zeiss Ultra55 before/after ferrofluid-raking showed varied results.

On some catalyst islands, the raking action was disastrous to the CNTs, leaving behind a debris field, with an outline shape similar to the shape of the catalyst island (i.e., round, square, etc). These debris fields (i.e., bits and pieces of broken-up nanotubes) were always shifted, relative to the islands, in the raking direction, with some evidence of alignment, as shown in Figure 2.

On other catalyst islands, we observed markedly different results. For example, Figure 3 displays SEM images of the same catalyst island before and after ferrofluid-raking. In the “Before” image, we see a nanotube forest wherein some of the nanotubes are in-focus and some are not. Many of the nanotubes had grown upward from the surface of the wafer. We believe that this growth was a consequence of the quartz boat that we used to hold the wafer chips in the furnace. Interestingly, at times, we could see certain portions of the carbon nanotubes moving about, depending on the settings of the SEM. In the “After” image, we notice that most of the carbon nanotubes that had grown out of the plane of the wafer were “scrubbed” away, leaving behind only those that were strongly pinned to the surface by Van Der Waals forces. In a few individual cases, we also noticed that a single tube had changed shape because an open loop in a tube that had been sticking-up into the air was folded down onto the surface as a result of the raking action.

In conclusion, we have successfully shown a mechanical interaction between the ferrous nanoparticles within a ferrofluid and CVD-grown CNTs.

Future Work:

By reducing the strength of the magnetic field and/or diluting the ferrofluid, it should be possible to accomplish CNT alignment without any destructive effects. It would also be interesting to try ferrofluids with different carrier liquids. Finally, we would like to try alignment of CNTs in liquid suspension. In other words, one would introduce CNTs directly into the ferrofluid and then see if a magnetic field could be used to control the alignment of the CNTs.

Acknowledgements:

I would like to thank my mentor Dr. Jay VanDelden for conceiving of this work and for his tireless effort. I would also like to thank Professor Sandip Tiwari, for allowing me to participate in his research group, the CNF staff, Professor Paul McEuen, for use of his furnace. Finally, I thank the Intel Foundation for their generous financial support and the National Nanotechnology Infrastructure Network REU Program for the summer.

References:[1] Ural, Li and Dai, “Electric-Field-Aligned Growth of Single-

Walled Carbon Nanotubes on Surface”, Applied Physics Letters, Vol.81, No.18, Oct.2002, pp.3464-3466.

[2] Liu, Casavant, Cox, Walters, Boul, Lu, Rimberg, Smith, Colbert and Smalley, “Controlled Deposition of Individual Single-Walled Carbon Nanotubes on Chemically Functionalized Templates”, Chemical Physics Letters, Vol.303, Apr. 1999, pp. 125-129.

Figure 2: Catalyst island with carbon nanotube debris field.

Figure 3: Before (left) and After (right) images.


MAT

ERIA

LS

Introduction:

MnAs is an attractive material for use in studies of ferromagnet/semiconductor heterostructures. Its room-temperature ferromagnetic properties and its ease of growth by molecular beam epitaxy (MBE) on arsenide semiconductors have motivated its use as a source of spin-polarized electrons in experiments in the field of spintronics [1]. In this work we present results on the fabrication and characterization of MnAs/GaAs nanowires. Freestanding nanowires with diameters down to 75 nm were fabricated and magnetic force microscopy (MFM) studies of nanowires with diameters down to 100 nm were conducted. Future experiments utilizing these nanostructures will provide information on the influence of geometrical confinement on the transport of spin-polarized electrons. Recent experiments on electron spin dynamics in InGaAs lateral channels have demonstrated one-dimensional electron spin dynamics in channels an order of magnitude greater than the electron mean free path [2]. The structures fabricated in this work could be used to do electrical spin transport experiments to further understand the effects of reduced dimensionality on spin-polarized transport in semiconductors.

Procedure:

The nanowires were processed from a thin film grown by MBE. The film consisted of a 25 nm thick layer of MnAs on top of 1.5 µm of n-GaAs (n = 5.25 x 1017 cm-3) grown on a semi-insulating GaAs substrate. Superconducting QUantum Interference Device (SQUID) measurements showed the MnAs layer to have a coercivity of approximately 200 Oe at 300 K.

The thin film samples were processed into freestanding nanowires using a two-step process. In the first step a pattern is defined on the sample surface via e-beam lithography. Using the negative maN-2403 e-beam resist as an etch mask, the pattern was transferred to the thin film via reactive ion etch (RIE), creating arrays of nanowires with diameters from 75 nm to 500 nm. The RIE was conducted using a Cl2 and Ar gas mixture with gas

Fabrication and Characterization of MnAs/GaAs Heterostructures for Studies of One-Dimensional Spin Transport

David ToyliPhysics, University of Minnesota

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigator: Professor David Awschalom, Physics, University of California, Santa BarbaraNNIN REU Mentor: Jesse Berezovksy, Physics, University of California, Santa BarbaraContact: [email protected], [email protected]

flow rates of 1 and 20 sccm, respectively, while the RF power was 50 W and the chamber pressure was 10 mT.

The magnetic properties of the MnAs nanowire caps were investigated with an Asylum MFP-3D atomic force microscope (AFM) using a magnetic cantilever tip. The AFM was used to take MFM images to characterize the ferromagnetic ground states of the MnAs caps. MFM images in an applied magnetic field were also taken with an Asylum research variable field module (VFM).

Results:

Freestanding nanowires with diameters down to 75 nm were successfully fabricated. Figure 1 shows a scanning electron microscope (SEM) image of a 75 nm diameter nanowire with a height of approximately 350 nm. The nanowire fabrication process was robust; typically 100% of the dots patterned during e-beam lithography were successfully etched into nanowires. Undercutting of the GaAs layer limited the nanowire heights—the greatest height to diameter aspect ratio achieved was approximately 6 to 1 using a Cl2-based RIE recipe.

Magnetic characterization of the nanowire caps done by MFM showed ferromagnetic behavior in MnAs caps with diameters down to 100 nm (the 75 nm nanowires were not measured because this array was not located during imaging). Figure 2 shows three nanowires of diameter 500 nm and height 350 nm. The image shows the MnAs

Figure 1: SEM image of a 75 nm diameter nanowire at 45°.


MATER

IALS

caps to be ferromagnetic. MFM scans also showed MnAs caps of diameter 500 nm to be either single-domain or multi-domain, and showed MnAs caps of diameters less than 500 nm to be only single-domain. Figure 3 shows an MFM scan of four 100 nm diameter nanowires. The MFM scan was done at a height of 275 nm above the sample surface to avoid contact with the nanowires. The large scan height results in poor resolution of the domain structure of the 100 nm MnAs caps. In order to obtain a better image of the MnAs caps, the nanowires were placed in a magnetic field along the MnAs easy axis large enough to reverse the magnetization of the MFM tip but not large enough to reverse the magnetization of the MnAs caps. Figure 3 shows that the fringe fields around the MnAs caps invert after the magnetization of the MFM tip is flipped. This response suggests that the MnAs caps are ferromagnetic with a coercive field larger than the applied field.

Future work will focus on using the MnAs/GaAs nanowires for studies of one-dimensional spin transport. Initial spin transport measurements could be done electrically at room temperature using a conductive atomic force microscope (CAFM) and a VFM. The CAFM measurements could be followed by low-temperature electrical measurements to provide a more complete understanding of spin transport in these

Figure 3: MFM images of 100 nm diameter nanowires imaged before (top) and after (bottom) MFM tip magnetization reversal. After switching the MFM tip magnetization the MFM image inverts, clearly demonstrating ferromagnetic behavior in the MnAs caps.

Figure 2: MFM image of 500 nm diameter nanowires overlaid on the AFM image (topography). The MFM image shows two single-domain MnAs discs (left, center) and one two-domain MnAs disc (right).

nanowires. Such low-temperature measurements would require nanowires processed so that the MnAs caps are contacted with a metal electrode. Processing these nanowires entails conformally coating the nanowires with a dielectric, planarizing between the nanowires using a spin-on dielectric like polyimide, and then etching to free the MnAs caps for metal evaporation. The purpose of the conformal dielectric coating is to structurally enhance the nanowires, as it was determined that the freestanding nanowires break at spin speeds greater than 3 krpm.

Acknowledgements:

David Toyli would like to acknowledge Professor David Awschalom and Jesse Berezovsky for their guidance and support in completing this project and acknowledge Shawn Mack and Roberto Myers for sample growth. This work was supported by the NSF, DARPA, and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Wolf et al. Science 294, 1488 (2001).[2] Holleitner et al. Phys. Rev. Lett. 97, 036805 (2006).


MAT

ERIA

LS

Abstract:Traditionally, heat transfer is modeled into three basic categories, conduction, convection, and radiation, all of which have their own equations that govern their procedures (Stefan-Boltzmann, Fourier, et al). Due to the evanescent effect found in optical research, it is speculated that at the nanoscale, heat transfer cannot be so sharply defined between the three categories as it is on the macroscopic scale. The focus of this project is to fabricate a chip that has a gap between two layers on the nanometer scale so that the transition between radiative heat transfer and conductive heat transfer can be studied.

This experiment never got through the fabrication step. There were various difficulties with the fabrication process that did not allow for the repeated production a successful chip. What will be outlined in this paper are the various attempts at making the nanogaps and the progress made in the fabrication of such devices.


The first method for making this gap was by spinning on e-beam PMMA-A4 resist onto a silicon-silicon dioxide wafer. Following this, micron sized holes were etched in the silicon dioxide so that a top layer of amorphous silicon (a-Si) could be deposited onto the wafer using plasma enhanced chemical vapor deposition (PECVD). Further, before attempting to remove the silicon dioxide release layer, a series of 4 µm wide trenches spaced 250 µm from each line, intersecting at 90 degree angles to each other were etched down to the substrate to allow the etchant to reach the release layer. Then various methods of undercutting were tested for the purpose of creating a gap between the two remaining layers of the chip. The results of the etches were then checked, first under a Nomarski interference microscope, and then a scanning electron microscope, if the situation warranted it.

The first attempts were used with a buffered oxide etch (BOE) of 20:1 and 6:1 ratios of hydrofluoric acid to ammonium fluoride, aqueous. This method of

Heat Transfer Through Nanogaps

Josef VeltenMaterials Science and Engineering, Illinois Institute of Technology

NNIN REU Site: Nanoscience @ UNM, University of New MexicoNNIN REU Principal Investigator: Dr. Kevin Malloy, Center for High Technology Materials, University of New MexicoNNIN REU Mentor: Ryan Martin, Department of Electrical Engineering, University of New MexicoContact: [email protected], [email protected]

undercutting had the disadvantage of taking roughly 48 hours to complete, and lifted off much of the top layer of a-Si leaving the bare substrate of silicon. The areas that did not have complete liftoff had buckling in the gap area, leaving the sample useless for heat transfer experiments. The second etchant used was a 49% by volume hydrofluoric (HF) solution. This solution etched through the release layer in under 9 minutes, but repeated the same issues as that of the BOE, namely buckling and liftoff of the top layer.

The second round of fabrication tests took the previous etchants and added surfactant into the etching liquid. The idea behind using the surfactants was to increase the wetability of the hydrophobic silicon and reduce the surface tension of the liquid to help with the speed of the etching and to help with the problem of bubbling, as by reducing the surface tension, the size of the bubbles would decrease. For the HF dip, various concentrations of an industrial standard detergent, FL-70, were tested on the chips. For the BOE etch, a premixed superwet 6:1 formula was used. The results from this round of etchant tests gave limited success with one of the HF/detergent mixtures. One of the patterns of the nanogap survived the etching process intact. However, while attempting to cleave the sample to look at it under the SEM, edge on, the cleave destroyed the nanogap that was made. Further attempts using the same formula and method did not yield similar results, so the original success was determined to be a fluke. However, progress was still made during this round of tests. Some of the top a-Si layers remained, but a certain amount of buckling still took place. The next round of fabrication tests would prove helpful to this problem.

The third round of tests were inspired by looking back at the manual of the PECVD machine that warned of using etchants when the temperature of the deposition was low, saying that residual strains on the a-Si would cause it to buckle and collapse when using etchants. The limitations of the PECVD that we had at UNM dictated that instead of depositing at a higher temperature, an extra step of annealing would be necessary to rid the a-Si of the


MATER

IALS

residual strain. Two procedures were tested in this step. The first method was to use the outgassing station of a high vacuum chamber in a slowly controlled temperature rise to the annealing temperature of 650°C, and a direct insert into a 700°C furnace. The furnace samples came out of the furnace with an immediately noticeable change in the surface-the top layer powdered off of the chip. After etching with surfactant the chip proved to have almost no buckling, but none of the nanogaps survived the process with either test.

Future Work:

The next round of tests are scheduled to take place after the submission date of this paper, and will involve a complete shift from e-beam lithography to optical lithography. The optical lithography was earlier disregarded as an avenue of approach for the reason that the size of the posts were required to be larger than 1 µm, which was the largest size post that we were originally willing to entertain. It is hoped that further work with the optical lithography will help define the smallest reasonable gap size for successful creation of the nanogap as well as speeding up the process for creating samples to work with. With the original method of using e-beam resist, it takes roughly 24 man-hours to go from start to etching per sample, the optical lithography promises to take less than half the time to go from start to finish. In conclusion there is much more work that needs to be done with the fabrication of the nanogap before the heat transfer can be studied readily.

Acknowledgements:

Thank you to the National Science Foundation and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.


MAT

ERIA

LS

Abstract:Silver nanocubes 30-50 nm in diameter have been synthesized using a polyol process in which silver nitrate is reduced by ethylene glycol in the presence of a capping agent, poly(vinylpyrrolidone) (PVP). A ligand exchange reaction was used to replace the PVP with another capping agent, allowing the nanocubes to be soluble in chloroform. Oleylamine, oleic acid, and decane-thiol were among the ligands investigated. The silver cubes were then used as sacrificial templates to generate hollow gold nanocages using a galvanic replacement reaction during which the silver cubes were titrated with chloroauric acid. The use of different capping agents allows us to further understand the role of the ligand in the galvanic replacement reaction.

Introduction:

The properties and applications of metallic nanostructures are mainly determined by their size, shape, composition, crystallinity, and structure (solid versus hollow). Optical properties of nanomaterials are of importance for the detection and treatment of cancer, particularly a phenomenon known as surface plasmon resonance (SPR), which refers to the characteristic wavelength at which free electrons in a nanomaterial collectively oscillate and scatter/absorb an incident electromagnetic wave [1].

The ability to tune the SPR of gold nanocages, which are hollow nanostructures with porous walls, into the infrared region promises uses as both contrast agents for optical imaging in early stage tumor detection and as therapeutic agents for photothermal cancer treatment [2].

The Xia group developed a process to synthesize gold nanocages with smaller dimensions than previously reported using a galvanic replacement reaction between silver templates and chloroauric acid [3]. This galvanic replacement reaction had previously been done in aqueous conditions with polyvinylpyrrolidone (PVP). Here we extended the replacement reaction to an organic

Synthesis and Galvanic Replacement Reaction of Silver Nanocubes in Organic Medium

Kaylie YoungChemistry, Brown University

NNIN REU Site: Center for Nanotechnology, University of WashingtonNNIN REU Principal Investigator: Younan Xia, Department of Chemistry, University of WashingtonNNIN REU Mentor: Xianmao Lu, Department of Chemistry, University of WashingtonContact: [email protected], [email protected]

medium, chloroform, for Ag nanocubes passivated with various capping ligands.

Experimental Procedures:

Silver cube synthesis: Silver nanocubes were prepared using the polyol process, during which ethylene glycol serves as both the reducing agent and the solvent. After heating ethylene glycol in a vial in an oil bath for one hour, the silver precursor, the capping ligand (PVP), and an etching agent were added. The reaction was stopped when the solution turned a grayish-silver color and appeared to be opalescent, which took about 15-25 minutes. The vial was submerged in cold water and the samples were washed once with acetone and twice with water to remove the excess PVP and ethylene glycol. The resulting silver nanocubes are shown in Figure 1.

Ligand exchange reaction: Before the ligand exhange reaction was performed, a small amount of PVP was added to the silver nanocubes to ensure their stability in chloroform. The silver cubes were then transferred from water into chloroform by evaporating off the water using a vacuum chamber and adding chloroform. The desired ligand was added in excess and the solution was briefly stirred and then allowed to sit for one hour.

Figure 1: SEM of silver nanocubes.


MATER

IALS

Galvanic replacement reaction: A dilute solution of chloroauric acid (HAuCl4) in chloroform was added slowly to the newly-capped silver nanocubes using a syringe pump. The reaction, which was done at an elevated temperature, proceeded through a number of color changes and was stopped at a grayish color. The resulting gold nanocages, shown in Figures 2 and 3, were washed with ethanol and stored in chloroform.


A galvanic replacement reaction has been demonstrated as a means for preparing gold nanocages using silver nanocubes as sacrificial templates. Since the standard reduction potential of AuCl4

-/Au (0.99 V vs standard hydrogen electrode, SHE) is larger than that of Ag+/Ag (0.80 V vs SHE), silver is oxidized to Ag+ when gold ions (HAuCl4) are added to the silver nanocubes [3]. The SPR peak was significantly shifted toward the infrared region for the gold nanocages versus the silver nanocubes. This is evident in the absorption spectra of the nanocubes and nanocages in Figure 4.

The morphology of the resulting gold nanocages was found to depend strongly on the capping ligand of the silver nanocubes used in the galvanic replacement reaction. The capping agent oleylamine gave the best nanocages with the highest yield. PVP-capped silver nanocubes resulted in some gold nanocages, although they were often found in large clumps and in low yields. In the cases of oleic acid and 1-decane thiol, no nanocages were formed. The thiol was thought to have bound so strongly to the silver nanocubes that it prevented the galvanic replacement reaction from occurring, leaving

the silver nanocubes as they were before the reaction. The oleic acid was thought to have strongly etched the silver nanocubes before the galvanic replacement reaction took place, resulting in tiny spheres and amorphous clumps. Neither nanocubes nor nanocages were visible after the galvanic replacement reaction.

Interestingly some batches of nanocages formed with oleylamine had the 100 faces of the cube etched while other batches had the 111 faces etched. This is illustrated by Figures 2 and 3. The difference between the batches was the amount of time that passed between the ligand exchange reaction and the galvanic replacement reaction. This suggests possible control over the face of the nanocube that etches. The possibility of selective etching of nanocubes needs to be further investigated in future work.

Acknowledgements:

This work was supported by the NSF and the 2006 National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program. Special thanks to my mentor, Xianmao Lu, and the rest of the Xia group for their encouragement and advice.

References:[1] Chen et al. “Gold Nanocages: Engineering Their Structure for

Biomedical Applications”; Advanced Materials, 17, 2005 (2255- 2261).

[2] Chen et al. “Gold Nanocages: Bioconjugation and Their Potential Use as Optical Imaging Contrast Agents”; Nano Letters, 5, 2005 (473- 477).

[3] Sun et al. “Mechanistic Study on the Replacement Reaction between Silver Nanostructures and Chloroauric Acid in Aqueous Medium”; JACS, 126, 2004 (3892- 3901).

Figure 2: SEM of gold nanocages with

100 faces etched.

Figure 3: SEM of gold nanocages with

111 faces etched.

Figure 4: Absorption of silver nanocubes and gold nanocages.


MAT

ERIA

LS

Abstract:The main focus of this research group is on getting material characterization of nanomaterials through Fourier transform infrared spectroscopy (FTIR). Due to the fact that FTIR mainly gives reflection coefficient vs. wave number, the transmission line theory is used to produce the sample and sample holder, and to transform data from FTIR into dielectric constants. The focus of this project is to first design the structure of the sample and the sample holder so that transmission line theory can be applied. This is important since ascertaining the thickness of the sample holder that will be hold the sample is key in getting as close as possible to the actual dielectric constant. The second focus is to design a simple diaphragm made of silicon nitride. Since we already know the dielectric constant of silicon nitride, this test will allow us to see how close the dielectric constant from FTIR is to the official one. The third step is to use the diaphragm as a sample holder and drop cast a layer of Ge nanowires made by the UTA CHE department. Then we will measure the dielectric of the Ge nanowires utilizing the above methods.

Introduction:

The main point of this research is to find a simpler and different way to gather basic material characterizations, conductivity and dielectric constant, of nanomaterials, instead of the traditional way of using electrical contact

Finding Dielectric Constant of Nanomaterials Using Fourier Transform Infrared Spectroscopy

Caleb YuElectrical Engineering, California State Polytechnic University, Pomona

NNIN REU Site: Microelectronics Research Center, The University of Texas at Austin NNIN REU Principal Investigator: Professor Dean P. Neikirk, Department of Electric and

Computer Engineering, The University of Texas at Austin NNIN REU Mentors: Dr. Sangwook Han and Joo-Yun Jung, Department of Electric and

Computer Engineering, The University of Texas at Austin Contact: [email protected], [email protected]

to measure the current and voltage and calculate material characterization from the data measured. This is important because in order for a material to be able to be used in actual applications, its basic properties need to be known. After doing some research and discussion with Professor Neikirk’s group, this research group figured out that FTIR, combined with transmission line theory, can be used to calculate dielectric constant. The group also noted that FTIR measurements are more effective on materials that are very transparent or absorbing. Those types of materials are also good for making detectors and electromagnetic structures [1,3,4,7].

Strategy:

The strategy was to first perform FTIR measurements on silicon nitride, with a known dielectric constant, to make sure that the equations derived from transmission line theory work and produce the desired result. Then FTIR measurements would be performed on a nanomaterial, Ge nanowires. The structure of the material samples were designed according to the quarter wave transformer circuit from the transmission line theory. Figure 1 shows the circuit of the Ge nanowire sample. Figure 2 shows the equations used to calculate dielectric constant from the Ge nanowire sample. Figure 3 shows the side view of the structure of the Ge nanowire sample [1,5].

Figure 1 Figure 2


MATER

IALS


The procedure for part one was basic. The first few steps included depositing silicon rich silicon nitride on the waver and spinning resists onto the waver. The last few steps included mask alignment, develop, exposure, RIE dry etching, and KOH anisotropic wet etching. The end result looked like Figure 3 without the Ge nanowires. For part two, a silicon nitride sample was made. Then the Ge nanowire powder obtain from Prof. Korgel’s group was mixed with chloroform or IPA solution. The mixed solution was then drop cast onto the back side of the silicon nitride sample. FTIR was then used to collect reflection coefficient from the different samples [2].


Unfortunately, the FTIR machine was not equipped to measure the reflection coefficient required by the equations. Only Foster Miller, a private company working with Prof. Neikirk, had the FTIR machine that could measure the reflection coefficient. But even with the measurement from Foster Miller, the data only had the magnitude and not the phase. This was a problem because the equations required the reflection coefficient as a complex number. Thus we could only measure transmittance vs. wavenumber of the samples and compare them to the measurement from Foster Miller to check to see if the FTIR test is performed correctly and the trend of dielectric constants as a function of wavenumber. Figure 4 shows the results from the silicon nitride sample. The transmittance of all the samples should have been at the same height to confirm the trend. The fact that they were at different heights was caused by the different ration of silicon vs nitride. For the Ge nanowire sample, we were not able to drop cast an uniform layer of Ge nanowire. Thus the data was incoherent because FTIR does not work on uneven surfaces. Furthermore, doing construction, it was noted that Ge nanowires mixed with IPA are less likely to cause the silicon nitride sample acting as a sample holder to break [6].

Future Work:

First, a FTIR with the ability to measure reflection coefficient and phase is required to do further work. Second, a better sample holder needs to be developed to hold the nanowire sample without breaking easily. Third, better casting techniques need to be found so that an even layer of nanowires can be achieved. Lastly, other type of nanomaterials will be used to perform the same experiment to see if dielectric constants can be calculated and see which materials are good for the FTIR measurement.

Acknowledgments:

Thank you to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, NSF, UT Austin, Microelectronic Research Lab, Professor Dean Neikirk, Dr. Sangwook Han, Joo-Yun Jung, and Yoon Park, Prof. Korgel’s group, and Foster Miller for all their help and support.

References:[1] David M. Pozar, Microwave Engineering, Addison-Wesley

Publishing Company, Inc. (1990) 67-94.[2] James D. Plummer, Silicon VLSI Technology, Prentice Hall,

Inc. (2000), chapter 5, 9, and 10.[3] MEE Handbook of Analytical Methods, Fourier Transform

Infared Spectroscopy (2006), http://www.mee-inc.com/ftir.html

[4] Nuance, What is FT-IR? (2006), http://www.nuance.northwestern.edu/KeckII/ftir1.asp.

[5] Constantine A. Balanis, Advanced Engineering Electromagnetics, John Wiley & Sons (1989) 145-150.

[6] Foster Miller Company.[7] Professor Dean Neikirk, Professor, Cullen Trust for Higher

Education Professorship in Engineering, The University of Texas at Austin.

Figure 3

Figure 4

MECHANICAL DEVICES • NNIN REU 2006 Research Accomplishments page 94

MEC

HA

NIC

AL

DEV

ICES

Abstract:The objective of the project was to examine the strengths and limitations of two etching methods of silicon carbide (SiC). The methods being investigated were reactive ion etching (RIE) and photo-electro-chemical etching (PEC). Reactive ion etching of SiC was done using SF6/O2 and SF6/Ar gas. The gas ratios and etch rates were evaluated. Surface morphology characterization by atomic force microscopy (AFM) was also performed on samples using both methodologies.

Introduction:

Silicon carbide (SiC) possesses characteristics such as high mechanical hardness, chemical inertness, large band gap, high thermal conductivity, and electrical stability at temperatures well above 300ºC, making it very attractive for high temperature micro-electro-mechanical systems (MEMS) applications. In order to create a SiC MEMS, the SiC must be etched. Some of the same properties that make SiC desirable for MEMS also limit SiC to certain etching techniques.

Procedure:

All RIE was done in a reactive ion etcher. After using photolithography to deposit a test pattern, a 1,000Å mask of Ni was evaporated on top of 200Å of Ti in order to protect the areas where etching was not desired. The thin layer of Ti was used because of its adhesive properties which allowed a successful liftoff. The chamber pressure was constant at 10 mTorr, time was constant at 30 minutes, and RF power was constant at 400W. While the gas flow ratio was changed, the total gas flow into the chamber was 50 sccm. Testing was done using a combination of SF6/O2 from 0:50 to 50:0 ratios increasing by 5 sccm. The same was done using SF6/Ar.

Characterization of Etching Techniques on SiC for High Temperature MEMS Applications

Lawrence BazilleMechanical Engineering, Southern University

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard UniversityNNIN REU Principal Investigator & Mentor: Dr. Gary L. Harris, Director,

Howard Nanoscale Science & Engineering Facility (HNF), Howard UniversityContact: [email protected], [email protected]

All PEC etching was done using an inexpensive set up. Inside of a container, electrical contact was made between the sample and a counter electrode. A 10% concentration of HF was then added into the container covering the sample. A potential difference of 10V was then added. Since the 6H-SiC samples were n-type, the actual oxidation and reduction process could not take place until a UV light was introduced at 1000W.

Results:

Using RIE, we obtained etch rates as high as 900Å/min using SF6/Ar at a 10:40 sccm ratio. All SF6/Ar combinations yielded smoother surfaces compared to the SF6/O2 combinations. RIE also produced vertical sidewalls between the etched and unetched areas as seen in Figure 1.

PEC etching yielded etch rates as low as 350Å/min. The surface of the PEC etched SiC was very porous as seen in Figure 2. It was also fairly rough compared to the RIE surfaces.

Figure 1: Vertical sidewalls produced by RIE.

page 95 NNIN REU 2006 Research Accomplishments • MECHANICAL DEVICES

MEC

HA

NIC

AL D

EVIC

ES

Conclusions:

We found that the maximum etch rate was obtained using an RIE etch with the SF6/Ar combination at 10:40 sccm ratio. It was also determined that an reactive ion etch of SiC yields a smooth uniform surface compared to a rough, porous PEC etched SiC surface as seen in Figures 3 and 4. It was also observed that RIE etching of SiC produces vertical sidewalls, while PEC etching of SiC produces an undercut under the unetched area.

Acknowledgements:

I would to thank Dr. Gary Harris, Mr. James Griffin, Dr. Peizhen Zhou, Crawford Taylor, Tony Gomez, Ms. Nupur Basask, my fellow interns, and all of the staff and graduate students at the Howard Nanoscale Science and Engineering Facility. Also, many thanks to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Z. Y. Xie, C. H. Wei, L. Y. Li, J. H. Edgar, J. Chaudhuri, C.

Ignatiev, MRS Internet J. Nitride Semicond. Res. 4S1, G3.39 (1999).

[2] J. A. Powell, P. G. Neudeck, A. J. Trunek, G. M. Beheim, L. G. Matus, R. W. Hoffman, L. J. Keys, App. Phys. Lett., 77 (2000).

Figure 2, top: An illustration of the porous 6H SiC surface after PEC etching.

Figure 3, middle: Fairly rough 6H SiC surface after PEC etching.

Figure 4, bottom: Uniformly smooth 6H-SiC surface after RIE.


MEC

HA

NIC

AL

DEV

ICES

Abstract:In this work, we report the design, fabrication, and preliminary experiments to study textured surface ratchets for droplet transport. The surface ratchet utilizes hysteretic interaction between a surface and droplet to yield a net force. By fabricating micro-scale surface roughness in phase with one side of a droplet and out of phase with the other, the total hysteretic force will have direction. Experiments have been done to determine the energy limits where this net hysteretic force can result in droplet movement. Edge-tracking software was developed to track droplet edge movement and to choose the correct range of operating parameters. Vibrating droplets of varying sizes were tracked on a first generation of straight ratchets. Based on these preliminary experiments, we have designed and fabricated curved ratchets. Movement was observed for certain ratchet designs.

Introduction:

Ratchets have long been an area of great interest for physicists and engineers. Rectifying motion has proved to be a very useful method of providing energy. While lures of the perpetual Brownian-ratchet still fascinate many people, more practical advancements turn up regularly. Ratchets continue to become more important especially

Droplet Transport Using Surface Ratchets

Dane TaylorElectrical Engineering and Physics, University of Wyoming, Laramie, WY

NNIN REU Site: Center for Nanotechnology, University of WashingtonNNIN REU Principal Investigator: Karl F. Böhringer, Electrical Engineering, University of WashingtonNNIN REU Mentor: Ashutosh Shastry, Electrical Engineering, University of WashingtonContact: [email protected], [email protected], [email protected]

as the scale of operation continues to decrease. This occurs as surface energies begin to dominate physical relationships, and requirements for system energy supplies decline.

One field of increasing importance is droplet microfluidics. This new area has far reaching impact, ranging from lab-on-chip systems to opto-fluidic components. Within this area, methods of droplet movement have become an essential field of research. Ratcheting droplets is a very promising solution to the problem of continuous, predictable droplet motion.

Surface Ratchet Theory:

The surface energy of the substrate was varied in an asymmetrical, periodic pattern as shown in Figure 1. This was achieved by microfabricating pillars on which a sessile droplet would rest. Commonly referred to as a Fakir state, it is possible for a droplet to sit on top of the pillars, or any roughened surface, in an equilibrium state. Because the bottom of a droplet only comes into contact with the pillar tops, the surface energy of the substrate is dependant on this surface area of the pillar tops. Furthermore, the pillar top surface area can be varied by changing pillar diameter or pillar density.

It is useful to represent the surface energy of a rough surface as a ratio of pillar top surface area to total surface area, which we refer to as the roughness factor. Figure 2

Figure 1: A contour plot of the surface energy is shown for multiple ratchets.

Figure 2: Computation of the roughness factor is shown for a curved surface ratchet.


MEC

HA

NIC

AL D

EVIC

ES

shows how surface roughness is quantified for a curved surface ratchet. This representation of roughness is very useful because it is directly proportional to the substrate’s surface energy. It is also used to provide estimations of a droplet’s contact angles.

Surface Ratchet Operation:

The first requirement for operation is that energy must be transferred to the droplet to cause the droplet footprint to expand and contract, or breathe. We used vibration at droplet resonance to transfer energy to the droplet, although any method to invoke droplet footprint oscillation would suffice. As the droplet edges oscillated, the three phase contact lines interacted with the rough surface. Hysteresis occurs and resists edge movement; however, because one droplet edge is in phase with the surface ratchet and the other side is out of phase, each contact line experiences different hysteresis. A much stronger hysteretic force is possible on the front contact line than the back contact line.

Based on estimations relating surface roughness with advancing and receding contact angles, it was apparent that a much stronger hysteretic force opposes receding contact lines than advancing contact lines. Thus, it is theoretically possible to provide enough energy to advance a contact line, but not enough to allow it to recede after advancement. This is precisely what occurs on the front edge of the droplet with each oscillation. During each breath, the droplet pulls itself along the ratchet.

One very important requirement for droplet transport to occur is that a very precise amount of energy must be provided to the droplet. If not enough energy is provided, the droplet oscillates without its edges moving. This is known as the “stick” energy region. If too much energy is provided, the droplet edges oscillates with too much energy for hysteresis to play an important role. This is known as the “slip” energy region. Therefore, there is a small window referred to as the “slip-stick” energy region, in which hysteresis dominates droplet edge movement.

Results:

Motion was observed for certain ratchet designs, providing the proof of concept. However, we are only at the early stages of understanding and controlling surface energy ratchet operation. Further work is being carried out to quantify motion and to develop the efficiency of the ratchets. Investigation is under way into varying ratchet characteristics to ensure more dependable droplet transport.

Acknowledgements:

The author wishes to thank the University of Washington and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Pierre-Gilles de Gennes, F. Brochard-Wyart, D. Quere.

Capillarity and Wetting Phenomena: Drops, Bubbles, Pearls, Waves. Springer-Verlag New York, Inc. 2004.

[2] X. Noblin, A. Buguin, and F Brochar-Wyart, “Vibrated sessile drops: Transition between pinned and mobile contact line oscillations,” Eur. Phys J. E 14, 395-404, 2004.

[3] A. Shastry, M. J. Case, and K. Bohringer, “Directing Droplets Using Microstructured Surfaces,” Langmuir, 22 (14), 6161 -6167, 2006.

[4] N. Patankar, “On the Modeling of Hydrophobic Contact Angles on Rough Surfaces,” Langmuir, 19, 1249-1253, 2006.

Figure 3: The forces acting on the droplet contact line are shown. The inward radial force is a result of minimizing droplet surface area. This is opposed by the outward radial force of hysteresis.


MEC

HA

NIC

AL

DEV

ICES

Abstract:Ultrasensitive microcantilevers can act as precise sensors if one measures their nanomechanical responses in different environments. The focus of this project was to detect microcantilever deflection when an intelligent hydrogel, applied to the surface, swells at variable pH levels.

The hydrogel was patterned onto microcantilevers of various sizes using UV photolithography. When the cantilevers were soaked in different buffers, the swelling hydrogel induced surface stress and therefore the bending phenomenon. This deflection was quantified by observing the change in focus with a microscope. The deflection was evident, and differed accordingly at various pH levels.

Introduction:

With the advent of micro and nanotechnologies, miniature sensing devices have featured prominently in today’s leading research. Sensors with the capability of detecting chemical and biological analytes are particularly exciting, since they can bring us one step closer to futuristic ‘lab-on-a-chip’ devices which can theoretically discover imbalances and automatically rectify them. And in the more immediate future, sensor-based devices can also serve as useful external diagnostic tools. Microcantilevers, commonly found in atomic force microscopy, have excellent potential as sensors due to several notable advantages including: ultrasensitivity, ease of mass production, and low cost.

The microcantilevers utilized in this project were placed in arrays on a small 2 x 2 cm2 silicon microchip. The goal of this project was to develop a microcantilever-based biosensor with the ability to sense minute changes in pH. The design is outlined in the following: an intelligent hydrogel (a polymer which swells when exposed to water), tailored to swell and collapse upon exposure to high and low pH, respectively, was patterned directly onto the cantilevers using photolithography. Dependent upon the amount of swelling due to pH,

Microcantilever-Based Sensors

Amit VasudevBiomechanical Engineering, Stanford University

NNIN REU Site: Microelectronics Research Center, The University of Texas at Austin NNIN REU Principal Investigator: Dr. Nicholas A. Peppas, Chemical and Biomedical

Engineering, The University of Texas at AustinNNIN REU Mentor: Carolyn Bayer, Biomedical Engineering, The University of Texas at AustinContact: [email protected], [email protected]

surface stress would induce the tips of the cantilevers to nanomechanically deflect, or bend, certain amounts. A microscope would then be used to measure the amount of tip deflection, which if present, and in agreement with the pH level, would essentially establish the device as functional. A literal depiction of the sensor is shown in Figure 1.


A. Application of Hydrogel to Cantilevers: The first step of this experiment was to functionalize the cantilevers’ bare silcon surfaces (as seen in Figure 2)

Figure 1: Cantilever deflection induced by swelling hydrogel [2].

Figure 2: Cantilevers, rightmost dim. 10 x 100 µm.


MEC

HA

NIC

AL D

EVIC

ES

with the treatment of a 10% γ-methacryloxypropyl trimethoxysilane/acetone solution for two hours. The purpose of this treatment was to provide an adhesive organic/inorganic interface between the silicon and the hydrogel. The intelligent hydrogel itself was a 1 gram monomer solution made up of methacrylic acid and polyethylene glycol 200-dimethacrylate (80:20 MAA : PEG200DMA mole ratio) with photoiniator 2,2-dimethoxy-2-phenyl acetophenone (10 wt%). Two drops of this solution were next spun at 2000 rpm for 30 seconds directly on the chip. Utilizing UV free-radical photolithography and the corresponding photomask, we were then able to accurately polymerize the monomer on the cantilevers at a power of 2.25 mW for 2 minutes. The instrument used was a Karl Suss mask aligner. Once polymerization was completed, the cantilevers were soaked in DI water in order to remove any unreacted monomer.

B. Observation of Deflection Patterns in pH Range 2-10 Using Microscope: The first step in the second half of the experiment was to make up a series of standard McIlvaine buffer solutions (0.5M constant ionic strength maintained by addition of potassium chloride) ranging from pH 2 through 10, with increments of 0.2 in the 5.8-6.8 range (as the most linear deflection patterns are thought to exist here [1]). Using a glass slide, tape, a hollowed spacer, and a coverslip, we were able to create a well-like apparatus in which the cantilevers could be firmly affixed at the bottom. This apparatus was placed under a 50x Nikon Eclipse ME600 microscope, and 100 µL of each buffer solution was individually pipetted into the well, effectively allowing for the hydrogel coatings to be fully immersed and swell to equilibrium. Once equilibrium was reached, the present buffer was removed and the next buffer pipetted in. As swelling occurred, deflection of the cantilevers was quantified using the change in focal length. Measurements of this kind were taken at 5-minute intervals for 30 minutes, the total time allotted for equilibration of each pH value.


The average measurable deflection of three trials over the 5.8-6.8 pH range was 10.833 µm/pH. The results of trial 2 are shown in Figure 3 where the deflection was 10.5 µm. The most important data to derive from these three trials, however, is a sensitivity of 0.0923 ∆pH/µm. From our results, it can be concluded that deflection of the cantilever tips due to hydrogel swelling was certainly evident and established proof-of-concept. It is important to note, however, that after the three aforementioned trials, those later trials using the full range of pH values were not consistent in yielding consistent results, often resulting in hardly any deflection at all. Sources of error may be due to one of several factors: inaccuracy of focal length adjustment, damaging of individual cantilevers due to repeated physical motion, and non-precise ionic strength of buffer solutions. Further, more precise tests for this particular sensor would need to be conducted prior to commercialization.

Acknowledgements:

I would like to thank Dr. Nicholas Peppas for allowing me to work on an exciting project, Carolyn Bayer for guidance along the way, and Dr. Sanjay Banerjee for use of an excellent facility. I would also like to thank Dr. Rashid Bashir and Mr. Amit Gupta for fabrication of the cantilever chips. Lastly, I thank coordinator Amy Pinkston and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for funding this research.

References:[1] R. Bashir, J. Z. Hilt, O. Elibol, A. Gupta, and N. A. Peppas, Appl.

Phys. Lett. 81, 16 pp 3093 (2002).[2] J. Z. Hilt, Novel BioMEMS Sensor Device: Ph-Responsive

Hydrogels Integrated With Silicon Microcantilevers, pp 116.

Figure 3: Results of trial 2 show max deflection of 10.5 µm.

OPTICS • NNIN REU 2006 Research Accomplishments page 100

OPT

ICS

Abstract:Optimized model for quantum cascade semiconductor ring lasers is reported. Using OptiBPM computer program, light propagation in the straight section of a ring laser is simulated. The optimal structure must support the fundamental mode of the light and confine the propagating light both laterally and vertically. The optimal design depends on the etching depth, ridge waveguide width, and separation between the ridge waveguide couplers. The etch depth is kept constant at 1.295 µm and the optimal coupling is observed with ridge width of 7 µm and 16 µm separation between the centers of the ridges.

Introduction:

Previous research on optoelectronic integrated circuits (OEIC) was based on quantum well (QW) and quantum dot (QD) semiconductor ring lasers [1]. OEIC based on semiconductor ring lasers are studied as potential monolithic ring laser gyros or rotation sensors. These devices are more durable compare to traditional gyroscopes because they have no moving parts.

Current research is being conducted on quantum cascade (QC) semiconductor ring lasers. Quantum cascade semiconductor lasers are unipolar semiconductor devices that emit mid-infrared light. They are advantageous in that they use less energy and the wavelength of the light can be easily tuned, and has a longer wavelength.

Purpose:

The purpose was to design mid-infrared ridge-waveguide directional couplers to be used as part of optoelectronic integrated circuits (OEIC) based on quantum cascade semiconductor ring lasers. A successful design was when the height and width of a single ridge are optimized to support the fundamental mode in the transverse and lateral direction. Figure 1 shows the height of the layers and the refractive indices of the quantum cascade structure.

Design of Mid-Infrared Ridge-Waveguide Directional Couplers by OptiWave Simulation

Luxue Rose DengPhysics, Hamilton College

NNIN REU Site: Nanoscience @ UNM, University of New MexicoNNIN REU Principal Investigator: Dr. Marek Osinski, Electrical & Computer Engineering, University of New MexicoNNIN REU Mentor: Dr. Gennady Smolyakov, Research Asst Professor, Electrical & Computer Engr, University of New MexicoContact: [email protected], [email protected]

In the real device directional couplers are adjacent to the straight section part of the ring laser. The energy coupled into the directional couplers indicates how the light is propagating in the laser. For the ring laser to function properly, less than 10% of the energy can be coupled into the directional coupler. The straight section part of the ring laser is 2 mm and 4 mm. The simulation of the couplers shows the coupling length, which is the distance it takes the energy to transfer from the ring laser to the directional coupler.

Experiment:

OptiBPM Designer was used to define the ridge-waveguide structure. In the program the appropriate size of the wafer and the material it consisted of was defined. Then the refractive indices of the layers of cladding, substrate, active region, and channels (ridges) were defined. The input plane, which dictates at what position the light starts propagating, was inserted. Another part of the program, Mode Solver, gives the number of modes in the system. Modes Solver was used to see if the mode was confined in the structure. The simulation parameters defined variables such as the wavelength of the light, the number of points in the mesh (resolution of the simulation), and the number of points displayed at the end of the simulation.

Figure 1: Layers of quantum cascade structure.

page 101 NNIN REU 2006 Research Accomplishments • OPTICS

OPTIC

S

The etching depth of the structure was kept constant at 1.295 µm, which meant there was 0.2 µm of cladding above the active region. The ridge-waveguide structures optimized for QW and QD laser structures did not confine the mode well laterally for the new QC laser structure because of the longer wavelength (3.6 µm) of the light in the QC structure. At 1.295 µm etching, the ridge width was increased by increments of 1 mm from 5 µm to 10 µm. Then for each structure, the Mode Solver found the modes the structure supported. The results were used to determine the ridge parameters that would be the best to design the directional couplers. Once the directional couplers were designed the OptiBPM Simulator was used to determine the coupling length.

Results/ Conclusion:

The ridge width that best confined the fundamental mode was 7 µm. The ridges of 8 µm width and above showed that the second order mode became well confined under the ridge (Figure 2). The OptiBPM Simulator was used for 10 mm of propagation length with 7 µm ridge

width and 16 µm separation between the centers of the ridges (Figure 3 and 4). As stated before, the QC ring lasers were designed with a straight section of 2 mm and 4 mm long. As the results of the simulation show, at 2 mm and 4 mm of the propagation distance, 2.5% and 10.3% of the energy was coupled respectively. Thus, this simulation showed the optimal coupling for 2-mm long straight section of the coupler. The coupling for the 4-mm long coupler can still be optimized by increasing the separation between the adjacent ridge waveguides.

Future Work:

Produce the masks needed for QC semiconductor ring laser fabrication. Fabricate the OEIC based on QC ring laser and test it to see if it is optimized for operation in mid-IR spectral range.

Acknowledgements:

Thanks to Professor Marek Osinski, Gennady Smolyakov, National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, National Science Foundation, Department of Defense, University of New Mexico, and The Center of High Technology Materials.

References:[1] Hongjun Cao; Chiyu Liu; Hai Ling; Hui Deng; Benavidez, M.;

Smagley, V.A.; Caldwell, R.B.; Peake, G.M.; Smolyakov, G.A.; Eliseev, P.G.; Osinski, M. In Applied Physics Letters, 24 Jan. 2005, vol.86, no.4, pp. 41101-1-3 : AIP, Journal Paper. (AN: 8275012).

[2] Hongjun Cao; Hai Ling; Chiyu Liu; Hui Deng; Benavidez, M.; Smagley, V.A.; Caldwell, R.B.; Peake, G.M.; Smolyakov, G.A.; Eliseev, P.G.; Osinski, M. In IEEE Photonics Technology Letters, Feb. 2005, vol.17, no.2, pp. 282-4 : IEEE, Journal Paper. (AN: 8269905).

Figure 2: Second order mode with quantum cascade structures with 1.295 µm etching, but different ridge widths.

Figure 3: Energy coupling in the active region of two ridge waveguide couplers.

Figure 4: OptiBPM 3D Isotropic Simulator. Simulation of 10 mm of the 1.295 µm etching, 7 µm ridge width, 16 µm gap between.


OPT

ICS

Abstract:Complementary metal oxide semiconductor (CMOS) compatible photonic devices hold great promise for the future of information technology, because optical interconnections have many advantages over electronic interconnections, including less delay and less power consumption. Silicon nanocrystals, a CMOS compatible photoluminescent material, can be fabricated by RF magnetron sputter deposition and annealing. In this project, a sample was made with a gradient of Si in a SiO2 matrix. Various annealing processes were done, including high temperature annealing, forming gas annealing (FGA), and high-pressure water vapor annealing (HWA). Photoluminescence (PL) was measured as a function of Si concentration to determine optimal light emission conditions. A peak shift in the PL intensity was observed, showing that the peak wavelength depended on the Si nanocrystal size. From these results, the conditions for optimal PL can be determined.

Introduction:

Although silicon is the only material used to fabricate CMOS processors because of its good electronic properties, it does not emit light. The nature of the indirect bandgap of bulk silicon does not allow radiative recombination of electron-hole pairs. Instead, they recombine nonradiatively, emitting phonons. Previous research groups have studied porous silicon [1] and silicon nanocrystals [2] as a possible light emitting source. The small size of a silicon nanocrystal (nc-Si) causes its wave function to be expanded in momentum space, due to the Heisenberg Uncertainty Principle, increasing the probability that an electron-hole pair can recombine radiatively and emit a photon. In this project, to find the optimum fabrication conditions for light emission, a non-uniform layer of nc-Si in a SiO2 matrix was fabricated across a substrate by the co-sputtering of Si and SiO2 and by annealing under various conditions. The variation across the substrate of light emission, layer

Photoluminescence of Silicon Nanocrystals Fabricated by Sputter Deposition and Annealing for Photonic Applications

Jenna HagemeierPhysics, Northwest Nazarene University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford UniversityNNIN REU Principal Investigator: Prof. Yoshio Nishi, Electrical Engineering, Stanford UniversityNNIN REU Mentor: James P. McVittie and Hiroyuki Sanda, Electrical Engineering, Stanford UniversityContact: [email protected], [email protected]

thickness, refractive index and Si composition were measured by photoluminescence (PL) ellipsometry and x-ray photoelectron spectroscopy (XPS), respectively.


We co-sputtered Si and SiO2 onto a 4ʺ silicon substrate without rotation so that an across substrate gradient of Si in SiO2 was achieved.

The samples were then annealed using the following processes: high temperature annealing at 1100°C for 1 hour, the same high temperature annealing followed by forming gas annealing (FGA) at 400°C for 30 minutes, and the same high temperature annealing followed by high-pressure water vapor annealing (HWA) at 26 atm and 260°C for 5 hours [3]. In this paper, positions on the wafer were assigned numbers 0-10 in 1 cm intervals, starting from the Si richest edge. A semiconductor laser with the wavelength of 532 nm was used for the PL measurement that ranges from 600 nm to 1100 nm.


The estimated excess Si in the sample from the XPS and the ellipsometric measurements ranges from 3% to 35%. The film thickness varied from 900Å to 1300Å across the wafer.

The PL intensity has its maximum at position 6, as shown in Figure 1. As the Si concentration increases, the size and the number of the nanocrystals increases (position 10-7), causing an increase in the PL intensity. This trend continues until a maximum point is reached, corresponding to 3.72% excess Si, a refractive index of 1.55, and a deposition thickness of 960Å. As the Si concentration increased beyond this critical point, the PL intensity decreases (position 5-2). This is probably because the nanocrystals are so close each other that electrons are becoming less confined. FGA enhances the PL intensity, which clearly suggests hydrogen in the forming gas is terminating dangling bonds. As a result,


OPTIC

S

the FGA decreases the number of interface states and increases the probability of photon emission. HWA does not significantly improve the PL intensity suggesting H from this source is less effective.

Figure 2 shows the peak wavelength as a function of position. Positions with higher silicon concentrations have longer peak wavelengths for all annealing conditions. This clearly indicates that the nc-Si size increases with silicon concentration since bandgap energy is inversely related to the nc-Si size. The FGA sample shows no significant peak shift. On the other hand, the HWA sample shows a significant peak shift to shorter wavelengths by 30-50 nm, which strongly suggests a smaller nanocrystal size due to oxidation by oxygen in the water vapor. The nc-Si radius that is estimated by peak wavelength at position 6 is around 0.4 nm after the high temperature annealing and the FGA, while it is reduced to around 0.3 nm after the HWA [4]. These results suggest the conditions for optimal light emission, which will help to develop Si photonic devices. The technique to fabricate a gradient nanocrystalline layer across a wafer is applicable to other materials and will be useful to further optimize the fabrication conditions.

Future Work:

Future work on this project includes measurements to quantify the Si concentration more accurately and to determine the Si nanocrystal size. We also fabricated Ge nanocrystals in SiO2 and SiGe nanocrystals in SiO2

[5]. Light emission in IR range suitable for optical communication as well as visible range could be expected from nc-Ge and nc-SiGe.

Acknowledgments:

I would like to thank Prof. Yoshio Nishi for allowing me to work in his research group, and Hiroyuki Sanda and Jim McVittie for mentoring me on my project. I also thank CIS, NSF and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for their support.

References:[1] Cullis, A., Canham, L., and Calcott, P., J. Appl. Phys., Vol. 82,

No. 3, 909 (1997).[2] Negro, L., Yi, J., Michel, J., Kimerling, L., Chang, T.,

Sukhovatkin, V., and Sargent, E, Appl. Phys. Lett., Vol. 88, 233109 (2006).

[3] Gelloz, B. and Koshida, N., J. Appl. Phys., Vol. 98, 123509 (2005).

[4] Iacona, F., Franzo, G., and Spinella, C., J. Appl. Phys., Vol. 87, 1295 (2000).

[5] Takeoka, S., Fujii, M., Hayashi, S., and Yamamoto, K, Phys. Rev. B, Vol. 58, No. 12, 7921 (1998).

Figure 1: PL intensity vs. position for all annealing conditions.Figure 2: Peak wavelength vs. position

for all annealing conditions.


OPT

ICS

Abstract:Single walled carbon nanotube field effect transistors (SWNT-FETs) have been the subject of much research over the last decade due to their unique electrical properties and small dimensions. In this project, we optimized suspended SWNT-FET structures to provide optimal conditions for small-diameter tube growth (< 1.5 nm), and to provide an even higher performance transistor. We also tried different SWNT growth/application methods, so as to understand which method would best fit a specific device.

Introduction:

Carbon nanotubes are a few nanometers in diameter, and can serve as metallic or semi-conducting materials. These are key qualities that have caused the SWNT-FET to be of rising interest. Previous groups have been successful in fabricating these SWNT-based field effect transistors, but the full potential of the device has not yet been realized. Our research focused on the improvement of these SWNT-FET structures and nanotube application methods.

Previous structures performed basic photolithography to pattern catalyst islands onto silicon oxide, so as to have a foundation of growth. The silicon oxide is grown on top of a highly doped P-type substrate used for electrostatic doping of the SWNT. A catalyst (cobalt acetate) is spun onto the substrate, and carbon nanotubes are grown from the catalyst. The idea is to get SWNT’s to grow from on end of the chip to the other. Two metal electrodes are deposited on top of the catalyst islands to provide electrical contacts for the SWNT. Another structure that is still being researched is the suspended carbon nanotube structure. The entire device is fabricated first, the oxide is etched down in the center to provide a trench, and tubes are grown and suspended over the trench. The nanotube is grown by laying the catalyst, performing the growth, and again relying on an end to end connection of the nanotube which lies on top of the metal electrodes. The metal used on the electrodes is tungsten and platinum. The local gate is also tungsten and platinum. To keep the

Suspended Carbon Nanotubes for Opto-Electronic Devices

Anthony SandersElectrical Engineering, Prairie View Agricultural and Mechanical University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford UniversityNNIN REU Principal Investigator: Prof. Hongjie Dai, Chemistry Dept, Stanford University NNIN REU Mentor: Xinran Wang and Yuerui Lu, Chemistry, Stanford University Contact: [email protected], [email protected]

source, drain, and gate metals from shorting out, a layer of nitride is grown on top of the oxide, and the oxide is wet etched to produce an undercut for the trench as shown in Figure 1. The light emission efficiency from these devices proved to be higher than the light emission from the devices where the SWNT is lying on the substrate. However, such suspended structure is only good for large diameter tube (> 1.5 nm) growth, as it’s difficult to grow small diameter tubes (< 1.5 nm) directly on top of metal. The wavelength that the nanotube emits is directly related to its diameter.

Our research not only focuses on optimizing the SWNT-FET device, but also growing smaller diameter tubes (< 1.5 nm) so that the wavelengths emitted will be in our detection range.


We fabricated a device very similar to the old suspended devices. A local gate was deposited into the trench. Instead of patterning our catalyst islands on top of the metal, we used e-beam lithography to dig a hole in the electrode down to the nitride as shown in Figure 2. The catalyst was then patterned on top of the nitride in an effort to produce small-diameter nanotubes.

The next device uses basic lithography to lay down four electrode pads on silicon. Using e-beam lithography, stems were etched onto the surface, one extending from each electrode. Two stems would lay 60 nm apart. About

Figure 1: Suspended device structure.


OPTIC

S

50 nm of gold was deposited on the stems to provide elevation. The idea was to simply stretch a tube across the stems to create a connection by spinning the tubes on the substrate.


Past research found that the metal/catalyst reaction causes small-diameter nanotube growth to be extremely difficult. Thus, the purpose of the device in Figure 2 was to construct the catalyst island as far from the metal as possible. Figure 3 is Raman spectrum data. The G-band occurs at 1600 cm-1, and it’s related to the amount of amorphous carbon material that’s present at the designated area. The tallest peak is silicon. The

radial breathing mode (RBM) peak relates to the yield of carbon nanotubes. Our results showed that we have a lot of carbon nanotubes lying on the catalyst, which is the area of interest that this particular data was extracted from. By taking 250 and dividing it by the RBM number, an estimate of the SWNT diameter can be obtained. We achieved a diameter of 1.1 nm which constitutes as a small diameter tube. Unfortunately through AFM imaging, we saw a very rough nitride surface and no tubes extending from the catalyst, causing the device to be without any connections.

The device highlighted in Figure 4 contained lift-off problems. Leaving a device in acetone for 2-3 hours is a normal lift-off procedure. With this device, sonication was absolutely necessary. This was in part due to the thickness of the metal as well as e-beam exposure and development issues. After many attempts at lift-off, we were able to fabricate a successful gold stem device.

Conclusions and Future Works:

We believe that the device in Figure 2 underwent a nitride/SWNT reaction that obstructed the growth of the SWNT’s. This reaction will undergo further investigation. The next step for the device in Figure 4 is to have SWNT’s spun on to its surface. Electrical and optical data can be taken once a connection is made on the device.

Acknowledgements:

The author wishes to thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Javey, A; Guo, J; Farmer, DB; Wang, Q; Wang, DW; Gordon,

RG; Lundstrom, M; Dai, HJ, “Carbon nanotube field-effect transistors with integrated ohmic contacts and high-κ gate dielectrics” Nano. Lett., 2004; v.4, no.3, p.447-450.

Figure 2: Suspended device with the catalyst island patterned on top of the nitride.

Figure 3: Raman spectrum data of the catalyst island. Figure 4: AFM image of a good lift off between the two stems.


OPT

ICS

Introduction:Metallic nanoparticles have become of interest for a wide

variety of applications areas due to their optical properties. These properties include optical surface resonance, localization, confinement, enhancement of electric fields, and resonance wavelength tunability as a function of geometry. The main goal of this study is to elucidate the viability of silver nanorings in achieving these optical characteristics. Silver having less optical loss than gold is expected to achieve better results. An additional goal of this study is to achieve the resonant wavelength redshift from visible towards infrared wavelengths often observed with gold nanoshells and gold nanorings when geometrical parameters are tuned. The geometrical parameters under consideration consist of varying the inner ring radius as compared to the outer ring radius and fabricating nanorings with subwavelength dimensions. To obtain subwavelength spatial dimensions the nanofabrication process includes e-beam lithography at high resolution. Preliminary results from nanofabricating the silver nanorings and optical analysis are presented.

The physical phenomena we expect to obtain from the silver nanorings involve surface plasmon resonances. In metals, the charge density undergoes periodic oscillations creating dipolar and multipolar patterns of electric charges. This in turn results in electric field enhancement in subwavelength regions near the top and bottom rims outside of the rings.

It has been shown with gold nanorings [1] that the extinction peaks are in the near-infrared region with peak wavelengths affected by geometry, the ratio of the ring wall thickness to the radius of the ring. As this ratio decreased, from 0.22 to 0.16 for the gold nanorings of [1], the resonant wavelength redshifts by 400 nm.

The importance of metallic nanoparticles is evident in the applications they contribute to. One application area is ultra-sensitive single molecule detection [2]. Cancer therapy has been under development and consists of applying gold nanoshells to treat cancer [3]. Another application concept involves the idea that materials on their own rarely resonate at the near-infrared or infrared wavelengths [4] and that the ability to tune to these wavelengths by changing geometry opens new application possibilities.

Silver Nanorings: Nanofabrication and Optical Properties

Julie SternPhysics and Chemistry, Stony Brook University

NNIN REU Site: Center for Nanoscale Systems, Harvard UniversityNNIN REU Principal Investigator and Mentor: Prof. Ken Crozier, Division of Engineering

and Applied Sciences, Harvard UniversityContact: [email protected], [email protected]

Materials and Methods:Silver nanoring arrays are made using a nanofabrication

protocol involving e-beam lithography and thermal evaporation of metal.

Pattern masks are designed within AutoCADLT 2005 and DesignCAD. Nanoring arrays were made to be 100 µm x 100 µm. The outer diameter of a ring was set to be 250 nm. The center to center ring distance was set to 500 nm. In the pattern mask, the ring wall thickness was set to a nominal value, and the parameters of e-beam lithography were varied to achieve different ring wall thicknesses. A smaller set of rings with the same size grid had dimensions: 120 nm outer ring diameter, and 500 nm center to center ring distance.

Figure 1 illustrates the steps of the nanofabrication process: e-beam resist, polymethylmethacrylate, is spun onto an indium tin oxide glass slide and baked, e-beam lithography is performed using the JEOL 7000 F SEM, development uses methyl isobutyl ketone, oxygen plasma cleans out the residual resist, chromium and silver are then thermally evaporated onto the slide, followed by an acetone soak and sonic vibration “lift-off” procedure.

To achieve the larger rings, the center-to-center e-beam pixel writing distance had to be reduced. Additionally, to achieve the smaller rings, a smaller e-beam dosage and smaller beam current were used.

Figure 1: Nanofabrication protocol.


OPTIC

S

Transmission data is measured from the fabricated nanoring array sample. Visible light from a xenon lamp irradiates the silver nanoring array head on. The light transmitted by the nanoring is collected into an optical fiber and into an optical spectrum analyzer. The data is normalized by incident light. A dip in transmission corresponds to a peak in absorption.

The SEM images were taken with the Leo 982 field emission scanning electron microscope.

Results and Conclusions:The resulting silver nanoring array produced from the

nanofabrication procedure and SEM imaging are shown in Figure 2. For a sample ring, the outer ring diameter was measured to be roughly 386 nm with a ring wall thickness of 80 nm. Figure 3 illustrates a tilted view of the same sample.

Smaller rings were nanofabricated (Figure 4). Only a minute amount of ring deformity and interior residual appeared. For a sample smaller-sized ring, the outer ring diameter was measured to be roughly 187 nm. The ring wall thickness was measured to be approximately 33.4 nm.

Optical characterization demonstrated resonances in the near-infrared wavelength range. In addition, the results showed that the resonances could be tuned to different wavelengths by changing the ring thickness.

Hence, the nanofabrication process worked for silver nanorings. High resolution e-beam lithography was needed to produce even smaller rings. The rings absorbed resonances within the near-infrared region of wavelengths. Geometric tunability of the optical modes was shown.

Acknowledgements:Special thanks to my mentor, Prof. Ken Crozier, Division of

Engineering and Applied Sciences, Harvard University. This work was supported by the NSF and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program. Thanks to Kathryn Hollar, DEAS, Harvard. Special thanks to Yuan Lu, Jiangdong Deng, Ed Macomber, Ling Xie, Steven Sheppard, and David Lange of the Center for Nanoscale Systems at Harvard University. Thanks to Emre Togan.

References:[1] Aizpurua, J., Hanarp, P., Sutherland, D. S., Käll, M., Bryant,

Garnett W., García de Abajo, F. J., “Optical Properties of Gold Nanorings,” Phys. Rev. Lett. 90, 057401 (2003).

[2] Alivisatos, P., “The use of nanocrystals in biological detection,” Nature Biotechnology, Vol 22, No 1, Jan (2004).

[3] Loo, C., Lin, A., Hirsh, L., Lee, M., Barton, J., Halas, N., West, J., Drezek, R., “Nanoshell-Enabled Photonics-Based Imaging and Therapy of Cancer,” Technology in Cancer Research and Treatment, Vol 3, Number 1, February (2001).

[4] Oldenburg, S. J., Averitt, R. D., Westcott, S. L., Halas, N. J., “Nanoengineering of optical resonances,” Chemical Physics Letters, 288, 243-247 (1998).

Figure 2: Silver nanorings.

Figure 3: Tilted view of silver nanorings. Figure 4: Smaller silver nanorings.


OPT

ICS

Abstract:The goal of this project is to create neodymium doped yttria-vanadate crystals (YVO4) using the ceramic method of production for potential use as a lasing medium. The method involves creating neodymium doped yttria stabilized zirconia and reacting them with LiVO3 to produce nanocrystalline neodymium doped yttrium vanadate crystals (Nd:YVO4). This reaction is expected to create crystals with higher concentrations of Nd in YVO4 than bulk synthesis allows, which could potentially yield increased lasing efficiency.

Introduction:One of the main limitations to improving solid-state lasers

are the concentration limitations of the luminescent dopant in the lasing medium. The recent advancements in ceramic production of lasing media combined with the increased dopant levels achievable in nanocrystalline materials may provide a means to push the concentration limits.

This project uses a LiVO3 flux technique to produce nanocrystalline neodymium doped yttrium vanadate (Nd:YVO4) from a neodymium doped yttria-stabilized zirconia base in order to reach higher concentrations of neodymium than bulk synthesis allows. Such nanocrystals could then be combined and sintered into a single body.

Nanocrystalline Nd:YVO4 Lasing Media

Chris StoaferChemistry, California Polytechnic State University, San Luis Obispo

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigator: Prof. David Clarke, Materials Dept. University of California, Santa Barbara NNIN REU Mentor: Matthew Chambers, Materials Dept. University of California, Santa BarbaraContact: [email protected], [email protected]

Higher concentrations of neodymium in the lasing medium may yield many benefits for this type of solid-state laser. For example, an increase in the efficiency of the lasing media may occur, allowing for more powerful lasers or smaller crystals and therefore more compact lasers. This could lead to a greater potential for laser applications and cheaper laser production.

Procedure:This method required an yttria-stabilized zirconia (YSZ)

precursor doped with rare-earth ions for the reaction to create the rare-earth ion doped YVO4 (RE:YVO4) crystals. Eu, which luminesces in the visible and whose spectra can be readily used to monitor changes in crystal structure, was used as well as Nd. TheYSZ was made with a precipitation method by mixing quantified amounts of zirconium acetate solution, yttrium nitrate, and the desired rare-earth ion nitrate (either europium or neodymium). This solution was then slowly added to concentrated ammonium hydroxide so the mixture would precipitate into nanocrystals. These nanocrystals were then collected by vacuum filtration, combined into a pellet and sintered.

This pellet was then placed into a platinum crucible with a proportion of LiCO3 and V2O5, so that when they melted they

Figure 1: Eu:YVO4 luminescence scan.

Table 1: ICP-AES results of HCl dissolution of neodymium flux products.


OPTIC

S

would form LiVO3 and CO2. This crucible was then heated to 1050ºC, where the LiVO3 melted and reacted with the crystal pellet to create the RE:YVO4 crystals, lithia, depleted zirconia and an excess vanadate. The water-soluble material (lithia and vanadate) was then removed with a water dissolution and the water-insoluble material was collected by vacuum filtration (RE:YVO4 is water-insoluble). A luminescence plot was then taken of this sample to determine what wavelength of light the rare-earth ion emits when energized (Figure 1). An SEM image with an energy dispersive x-ray elemental map were taken to determine what compounds were in the water-insoluble material. A portion of this material was then dissolved in concentrated HCl to dissolve the RE:YVO4. This solution was then analyzed with inter-coupled plasma-atomic emission spectroscopy (ICP-AES) to determine the concentration of the rare-earth ion in the crystal.

Discussion and Conclusions:The luminescence plot of a sample made with europium

showed significant peaks around 615 nm and 620 nm as seen in Figure 1. The spectrum is dissimilar to that seen in europium doped zirconias in that there is no peak observed at 580 nm, and the 5D0 ‡ 7F1 triplet centered at 590 nm in zirconia appears as two triplets centered at 585 nm and 595 nm; this result suggests that the rare-earth dopant is fully leached from the zirconia during the reaction. The SEM images (for example in Figure 2) showed a mixture of differently shaped nanocrystals and the energy dispersive x-ray results showed that YVO4 crystals had been produced along with zirconia crystals, and some of the vanadate failed to dissolve (shown in Figure 3). The ICP-AES showed that the percent of the rare-earth ion (in this case neodymium) in the HCl-dissolvable material (Nd:YVO4, but not ZrO2, can be dissolved in HCl) was around 13%, which is much higher than the normal percentage in bulk-synthesized lasing media which is about 1%. These ICP-AES results indicate that the rare-earth dopant becomes concentrated in the YVO4 crystals during processing, but since the YVO4 crystals were not isolated there may have been some rare-earth

ion leached from other phases in the sample, resulting in an over-estimate of dopant concentration in the YVO4 crystals. The completed analysis gave some initial results about the success of our method, but more work needs to be done in order for these analytical techniques to be more accurate, and lasing efficiency has not yet been assessed.

Future Work:The next step, to proceed in the analysis and development of

this method, would be the isolation of the RE:YVO4 crystals. More ICP-AES tests would then be run to determine the concentration of the rare-earth ion in the crystals. Following the ICP-AES analysis, the lasing efficiency of the crystals would then be measured and the method optimized to produce the highest lasing efficiency possible.

Acknowledgements:I would like to thank Matthew Chambers and Professor

David Clarke for their support and knowledge. I would also like to thank Vladimir Tolpygo for his SEM imaging expertise and the staff of the Materials Processing Lab at University of California, Santa Barbara. This project was support by the National Science Foundation and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] Erdei, S. Growth of oxygen deficiency-free YVO4 single crystal

by top-seeded solution growth technique. (1993) Journal of Crystal Growth. 134:1-13.

[2] Wikipedia contributors, ‘Neodymium doped yttrium orthovanadate’, Wikipedia, The Free Encyclopedia, 18 March 2006, 03:36 UTC, http://en.wikipedia.org/w/index.php?title=Neodymium_doped_yttrium_orthovanadate&oldid=44306517.

Figure 2: SEM of product mixture from Eu:YVO4 flux.

Figure 3: Energy dispersion x-ray of SEM sample.


OPT

ICS

Abstract:In this project we perform free-space characterization of stacked InAs quantum dot devices. A crossed polarizer and analyzer combination is used to determine the phase retardation/voltage relation for said materials. We use a span of pump wavelengths to analyze the wavelength dependence on modulation. Further calculations are carried out to determine the theoretical extinction ratio of such devices as part of a mach-zender modulator.

Introduction:

Mach-zender (MZ) devices are an ideal modulation source for communication networks at 1.3 µm and 1.55 µm. Superlinear electro-optic (EO) effects are a desirable feature in MZ modulators since their large second order EO coefficient will theoretically provide complete signal extinction at a small voltage. Quantum dot (QD) devices show promise for such applications in the 1.3 µm band.

Quantum Dot Modulators

Brendan TurnerPhysics, Brigham Young University

NNIN REU Site: Nanoscience @ UNM, University of New MexicoNNIN REU Principal Investigator: Diana Huffaker, Electrical and Computer Engineering, University of New MexicoNNIN REU Mentors: Manish Mehta, Electrical and Computer Engr, Ramesh Laghumavarapu, Physics; University of New MexicoContact: [email protected], [email protected]

Device Structure, Growth and Fabrication:

Stacked InAs/GaAs QD structures were grown by metalorganic chemical vapor deposition on an n-doped GaAs substrate [1]. Two layers of n-doped AlxGa1-xAs (values of x used were 0.35 and 0.6) were grown at 700ºC followed by a GaAs separate confinement heterostructure (SCH) layer. Six stacks of InAs quantum dots were grown at 520ºC with intermixed GaP strain compensating layers. Another GaAs cladding layer was followed by two p-doped AlxGa1-xAs layers and one layer of p-doped GaAs, all grown at 560ºC.

After growth, the photo-luminescence response was taken to test the active layer. TiPtAu is evaporated to form the top p-contact, followed by a masked mesa etch using BCl3 in an inductive coupled plasma reactive ion etcher. The mesas were etched to 200-300 Å above the SCH layer. Ge/Au/Ni/Au was then evaporated onto the backside of the substrate to form the n-contact and the wafer was cleaved into 2 mm stripes. The sample was then tested to determine its electro-luminescence (EL) response, mounted, and wire-bonded.

Experimental Setup:

Free space characterization of the InAs QD devices was performed with a probe wavelength of 1350 nm. This wavelength was chosen by red-shifting the peak EL wavelength to reduce absorption in the active layer.

Figure 1: InAs QD

laser structure.

Figure 2: Setup for measurement of QD modulator.


OPTIC

S

The probe laser was coupled into the active region using a 100X objective lens and the emitted light was gathered and collimated using a 40X objective. The collimated beam was measured using a broad-area IR detector combined with a chopper and lock-in amplifier (see Figure 2). The incoming beam was polarized at a 135º relative to the sample. This angle was chosen so as to give equal vertical and horizontal components of the E-field vector so the modulated light would be linearly polarized. An analyzer was placed after the sample at 45º to eliminate scattered and un-modulated light from reaching the detector.

Initial alignment was done with a HeNe laser aligned to follow the same path as the pump laser. Precise alignment was done using the pump laser and an IR viewing card as well as an IR sensitive CCD camera. The beam was focused on the top edge of the front facet using a beam splitter and the IR viewing card to look at the reflection from the facet. Care was also taken to couple the pump laser into one of the wire-bonded channels. The packaged sample was connected to a power supply and alignment of the second objective was optimized by applying a forward bias to the sample and maximizing the reading of the lock-in amplifier.

Modulation readings were taken by applying a voltage in the reverse bias to the sample using the power supply. Any modulation in the phase of the probe beam registered as an increase in transmission through the crossed polarizer/analyzer pair. Modulation was optimized by applying a reverse bias and adjusting the height of the sample to obtain the maximum amount of light coupled into the quantum dot layer. Various lock-in measurements were taken from 0 to 7 Volts.

Results and Calculations:

Calculations of the phase retardation were performed by averaging the readings at each voltage and subtracting

the background noise. The maximum lock-in reading was assumed to be a 90º phase retardation. All other readings were divided by this value and the square root and arcsine were taken. While previous research has shown a super-linear EO effect for similar InAs QD structures all results were shown to be linear (see Figure 3). This relation was fitted to a linear trendline and this equation was used to calculate the extinction ratio of an ideal MZ modulator employing an InAs QD active layer. First an intensity curve was calculated by adding two sine waves with the phase of one modulated by the phase reduction/voltage relation. This was squared and averaged to determine the voltage for maximum extinction. Also the ratio of full intensity to the intensity of two 180º out of phase was calculated to determine the extinction ratio (see Figure 4). Based on these calculations a theoretical mach-zender modulator could achieve full extinction at approximately 9 V for a 2 mm sample.

Further experiments need to be conducted to determine if the EO linearity is dependent on the probe wavelength and what amount of red-shift will maximize the EO coefficients.

Acknowledgments:

Many thanks to Manish Mehta , Ramesh Laghumavarapu, Diana Huffaker and everyone in the Huffaker group. Also much is owed to Melanie-Claire Mallison, Lynn Rathbun, and in general the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for organization and funding.

References:[1] J. Tatebayashi, N. Nuntawong, Y.C. Xin, P.S. Wong, S. Huang,

C.P. Huang, L.F. Lester, D.L. Huffaker, “Low threshold current operation of stacked InAs/GaAs quantum dot lasers with GaP strain-compensation layers,” International Conference on Indium Phosphide and Related Materials, 108-111, 2006.

Figure 3, left: Phase retardation vs. voltage.

Figure 4, right: Theoretical extinction ration for InAs QD

Mach-Zender modulator.


OPT

ICS

Abstract:This project aims to characterize the wavelength dependent transmission characteristics of a straight silicon-on-insulator waveguide coupled to a silicon-on-insulator ring resonator. The transmission loss of straight silicon waveguides was determined through Fabry-Perot loss measurements. Loss coefficients of 0.194 ± 0.037 cm-1 and 0.105 ± 0.08 cm-1 were obtained for waveguides of 1 µm and 1.5 µm widths, respectively. The ring resonator parameters, such as ring loss, ring-waveguide-coupling efficiency, quality factor, and finesse are determined by taking wavelength dependent transmission measurements for various ring radii, ring-waveguide-gap distances, and waveguide widths. For rings with a radius of 400 µm, a quality factor of up to 4.05 x 104 and finesse of 7.35 have been measured for a width of 1 µm and a Q of 2.59 x 105 with a finesse of 46.94 have been measured for a width of 1.5 µm.

Introduction:

Silicon ring resonators are becoming an increasingly interesting area of focus in silicon photonics. When light of specific wavelengths is coupled into the ring resonator, there can be a build up or cancellation of optical power due to interference between the light from multiple round trips within the resonator. This, in conjunction with silicon’s low loss coefficient, makes silicon ring resonators excellent passive low-loss filters with high quality factors.

Experimental Procedure and Device Fabrication:

The Fabry-Perot loss and ring resonator transmission measurements were made by launching light into the device under test from a tunable laser source, sweeping at wavelengths starting from 1550 nm, through a lensed optical fiber. The span and step of each sweep is determined by the calculated free spectral range so that each scan yields 6 periods and each period has a resolution of at least 250 points. At the opposite waveguide facet, the light is first collimated through an

Loss, Reflection and Transmission Measurement and Analysis of Silicon-on-Insulator Ring Resonators

Jason WangElectrical Engineering and Bioengineering, University of Pennsylvania

NNIN REU Site: Nanotech @ UCSB, University of California Santa BarbaraNNIN REU Principal Investigator: Dr. John E. Bowers, Electrical and Computer Engr, University of California at Santa BarbaraNNIN REU Mentor: Alexander Fang, Electrical and Computer Engineering, University of California at Santa BarbaraContact: [email protected], [email protected], [email protected]

80x microscope objective and the TM light is filtered out through a polarizing beam splitter. The TE light is then collected by the photodetector.

The waveguides used to form the devices were all rib waveguides with a height of 0.9 µm and a rib etch depth of 0.2 µm. The waveguides were fabricated on SOI in a standard CMOS fabrication facility. The rib waveguides were etched using Cl2/HBr/Ar to perform inductively coupled plasma etching. A thin layer of SiO2 was used as a hard mask.

Results and Conclusion:

Figure 1 shows the equation relating the loss coefficient to the waveguide length, reflectivity, and the ratio of the minimum to maximum intensity (ζ). Therefore if ζ is measured for two different waveguide lengths, then a system of equations will allow for the reflectivity and loss coefficient to be determined. For waveguides with a width of 1 µm, the average ζ ± SD is 0.277 ± 0.033 and 0.313 ± 0.026 for a length of 3.2 mm and 8 mm, respectively. This results in R = 0.330 ± 0.014 and an α = 0.194 ± 0.037 cm-1. A waveguide 1.5 µm wide has an average ζ = 0.298 ± 0.029 and ζ = 0.317 ± 0.027 for a length of 3.2 mm and 8 mm, respectively. This yields an R = 0.304 ± 0.02 and α = 0.105 ± 0.08 cm-1.

Figure 2 shows the schematic of the ring resonator with the corresponding theoretical equation relating the input and output power that is used to fit the data. The waveguide facets were AR coated to minimize the Fabry-

Figure 1: Equation for Fabry-Perot Loss Measurement [1].


OPTIC

S

Perot response of the straight coupling waveguide. The data was normalized by dividing out the response of an AR coated straight waveguide. A fit of the normalized data was made using the equation given in Figure 2 by varying and thereby extracting the propagation loss of the ring and the coupling efficiency of the waveguide. These variables allow for the calculation of the quality factor and finesse—both of which are descriptions of peak sharpness and the strength of the resonance of the cavity. Figure 3 shows the wavelength dependent transmission data and theoretical fit for a ring resonator with a 400 µm radius, 0.9 µm gap, and a width of 1.5 µm. This fit results in a quality factor of 2.59 x 105 with a finesse of 46.94, a group index of 3.6766 and a loss coefficient of 0.38 cm-1. Figure 4 provides a table summarizing the previously mentioned parameters for rings with a radius of 400 µm but varying waveguide width and gap distance. As expected, α for the rings was higher than that of the straight waveguide due to bend loss and the additional scattering. Furthermore, the 1.5 µm wide waveguides had a smaller loss coefficient than the 1 µm wide rings.

Finally, as the gap distance increased, the quality factor and finesse increased, showing stronger resonance and greater wavelength selectivity. This is expected, as a greater gap distance lowers the coupling coefficient between the ring and waveguide. This leads to a higher feedback coefficient and allows for stronger interactions between the light from multiple roundtrips in the ring.

Rings of smaller radii were tested but due to the high bend loss at these smaller radii (caused by shallow rib etch) the fit was unable to be made due to high noise even after normalizing the data. From the data, the ring response for the wavelength sweep was on the order of the AR coated Fabry-Perot response and noise of the experimental setup.

Acknowledgements:

I would like to thank Dr. John Bowers, Alexander Fang, Hyundai Park, and the rest of the Bowers Group for their help and support through this project, and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and NSF for this opportunity. Special thanks to Oded Cohen at Intel for device fabrication and Kim Callegeri at Intel for sample preparation.

References:[1] Tittelbach et. al. “Comparison of Three Transmission Methods for

Integrated Optical Waveguide Propagation Loss Measurement.” Pure Appl. Opt. 2 (1993) 683-706.

[2] Rabiei et. al. “Polymer Microring Resonators.” Optical Microcavities, 319-366.

Figure 2: Ring schematic and equations for theoretical fit, Q and F [2].

Figure 3: Ring measurements with theoretical fit showing wavelength dependent transmission. Figure 4: Table summarizing ring resonator parameters.

PROCESS & CHARACTERIZATION • NNIN REU 2006 Research Accomplishments page 114

PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:Conventional electron-beam lithography is done using a scanning electron microscope (SEM), with a resolution limit of ~ 10 nm [1]. However, there is continued need for higher resolution lithography.

The goal of this project is to investigate higher resolution electron-beam lithography using a scanning transmission electron microscope (STEM). In principle, the STEM has two main advantages: less scattering of incoming high energy electrons, and a smaller electron probe diameter. We have created 100 nm wide trenches in PMMA resist, which are promising early results. Reducing the exposure of the resist will likely give higher resolution.

Research and Development of Electron-Beam Lithography Using a Transmission Electron Microscope at 200 kV

Michael AdamsPhysics, University of North Carolina at Chapel Hill

NNIN REU Site: Center for Nanoscale Systems, Harvard UniversityNNIN REU Principal Investigator & Mentor: Dr. David Bell, Center for Nanoscale Systems, Harvard UniversityContact: [email protected], [email protected]

away, leaving the desired pattern in the substrate (see Fig.1).There are two common lithography methods. The first,

called photolithography, is shining UV light through a plate with the proper pattern engraved in it, exposing light-sensitive resist beneath in the same pattern. The limit of the smallest feature that can be produced (the resolution) depends linearly on the wavelength of light used. State of the art deep UV photolithography using light wavelengths of 193 nm has a resolution of ~ 50 nm [2]. However, for some applications, higher resolution is needed. Thus, the second common lithography method is electron-beam lithography.

Electron-beam lithography uses electrons instead of photons to expose a resist. Because electrons have a much smaller wavelength than the light used in photolithography, then electron-beam lithography has a much higher intrinsic resolution. There is also a difference in that the electron beam can be focused onto a substrate directly and controlled so it only exposes those areas which ought to be exposed, without needing a mask to block certain light from reaching certain areas. Conventional electron-beam lithography is done using an SEM. The SEM can give electrons energies up to about 30 KeV, and uses an electron probe with a diameter of a few nanometers. Due to scattering, however, the maximum resolution is about ~ 10 nm [1].

There are two main types of scattering—forward and back scattering. In forward scattering, the paths of the incoming electrons are deflected by the atoms’ coulomb potential into a cone-like trajectory. In backscattering, the path is deflected by an angle greater than 90 degrees, and the electrons go back to expose a much larger area of the resist than the area of the incoming electron beam (see Figure 2). The lower the electron energy, the more likely backscattering is to occur. The SEM has relatively low beam energy; thus, backscattering is an important problem. If two lines are written too close together, then the backscatter will end up exposing all the area in between. The TEM electron beam has a much higher energy, substantially reducing the backscatter. Thus, the TEM or its cousin, the scanning TEM (STEM), should have an inherently higher resolution.

In the late 1980s, STEM was tried for electron-beam lithography [3]. Early experiments with the STEM succeeded in producing patterns with resolution of ~ 10 nm [3]. Few groups have worked on this since, despite the availability of new electron-beam resists. It was our purpose to further explore

Introduction:Lithography is the creation of three dimensional structures

on a substrate that allows for transfer of a pattern to that substrate. First, a “resist”—a material sensitive to light or electrons—is deposited on a substrate. Certain areas of the resist are “exposed” by light or electrons, making the area more susceptible to a subsequent chemical treatment, called “development”. After development, the substrate has only certain places where it is still covered by resist. The substrate is etched or additional material is deposited (the remaining photoresist covers the area that is not meant to be etched or to have material deposited on it). Finally, the resist is stripped

Figure 1: Electrons exposing certain areas of resist, which development then eliminates.

Figure 2: Electron beam is scattered, exposing larger area of resist.

page 115 NNIN REU 2006 Research Accomplishments • PROCESS & CHARACTERIZATION

PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

this idea by using a new resist, a high accelerating voltage in a STEM, and a thin sample.

Experimental Procedure:We explored two different methods for writing with an

STEM. The first was a lithography setup, where we spun coated an HMDS adhesion layer and then a layer of PMMA 495 C3 positive electron-beam resist on a 100 nm thick silicon-nitride membrane, and then used a JEOL 2010 STEM with a Nabity pattern generator attached to expose the resist. We developed with MIBK 1:3 to clear away the exposed areas, and examined the patterns under the SEM. The idea behind using a silicon-nitride membrane is that this should reduce backscatter even more, thus potentially giving even cleaner, smaller lines.

The next method we tried was direct writing on a thin metal layer that was deposited on the 100 nm Si3 N4 membrane. The 200 keV beam energy of our STEM should have been enough to ablate the metal. We demonstrated this in principle by ablating holes in a Si3 N4 membrane without a metal layer. We then thermally evaporated 10Å of aluminum onto the Si3 N4 membrane. However, the metal did not adhere well enough to form a uniform layer; instead, the aluminum aggregated into clumps on the membrane surface. For better adhesion, we tried thermally evaporating 20 nm of chromium onto the membrane surface. This still clumped (see Figure 4), making pattern writing impossible.

Results and Conclusions:With our lithography setup, we have created some trenches

in the PMMA resist (see Figure 3). We would like to achieve finer resolution than that demonstrated (about 100 nm is the smallest line width), but this is an important first step. One area that needs to be examined more is the current dose used in the writing. We used doses between 1 and 16.5 nC/cm. However, given that some of our lines blended together, much lower doses may need to be explored.

In our direct writing setup, we need to reduce the aggregation problem, and achieve a uniform layer. There are several possibilities to explore. Different metals could potentially be thermally deposited in a much more uniform way, or a different process altogether may be more appropriate—for instance, sputter coating the metal on the Si3N4 membrane may give the metal atoms enough energy to stick in place on the membrane, rather than allowing some to aggregate.

Acknowledgements:Thanks to Dr. David Bell, Yuan Lu, Dr. Jiangdong Deng,

Stephen Shepard of CNS. Thanks also to Kathryn Hollar of the Harvard REU program, and Melanie-Claire Mallison of the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

References:[1] C. Vieu, F. Carcenac, A. Pepin, Y. Chen, et al, “Electron Beam

Lithography: Resolution limits and applications”, Applied Surface Science 164, 111-117, 2000.

[2] Richard J. Blaikie, David O.S. Melville and Maan M. Alkaisi, “Super-resolution near-field lithography using planar silver lenses”, Microelectronic Engineering 83 (4-9): 723-729 Apr-Sep 2006.

[3] A.N. Broers, A.E. Timbs, R. Koch, “Nanolithography at 350 KV in a TEM”, Microelectronic Engineering 9, 187-190, 1989.

[4] Jones GM, Hu BH, Yang CH, Yang MJ, Lyanda-Geller YB, “Observation of one-electron charge in an enhancement-mode InAs single-electron transistor at 4.2 K”, Applied Physics Letters 88 (19): Art. No. 192102 May 8 2006.

Figure 4: 20 nm chromium “clumps” on 100 nm Si3N4 membrane.

Figure 3: 100 nm trenches in PMMA resist on Si3N4 membrane.


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:Localized surface plasmon resonance (LSPR), a collective electron density oscillation found exclusively in metallic nanostructures, is a phenomenon that is of practical significance. The LSPR response of nanoparticles to changes in their surrounding dielectric environment may be exploited to use nanoparticle arrays as sensing platforms for biological or chemical sensors. This project focuses on the fabrication of such platforms using the method of nanoimprint lithography (NIL). NIL provides a way to produce significant areas of monodispersed nanoparticles of controlled size, shape and composition directly onto a wide range of substrates using a two-dimensional nanoblock array mold. Using this method of fabrication, we will study how the surface plasmon resonance of our fabricated Ag and Au nanoparticle arrays is affected as their composition as well as dielectric environment changes.

Introduction:Localized surface plasmon resonance has been studied

extensively over the past decade due to its utility as the backbone of many photonic technologies. Using spectroscopy, our LSPR biosensors will perform refractive index sensing by transducing changes in the surface refractive index into wavelength shifts of the LSPR extinction maximum.

The extinction maximum that we witness is the result of light absorption and scattering that occurs as light is shined onto our patterned sensing platforms. The electric field component of the incident light interacts with the particle electrons, causing them to collectively oscillate. LSPR is sensitive to the size, composition and shape of the nanoparticles as well as their orientation, spacing and dielectric environment. An abundance of nanofabrication techniques have been employed to produce the desired nanostructures utilized for LSPR biosensors with varied degrees of success as measured by the previously mentioned parameters that affect LSPR. Nanoimprint lithography is a very powerful method that allows for great particle fabrication control. Other alternative methods such as nanosphere lithography possess inherent fabrication limitations that allow for little particle orientation control, long range order, as well as particle geometric variation.

Fabrication of Metallic Nanoparticle Arrays

Christine ChinMechanical Engineering, Massachusetts Institute of Technology

NNIN REU Site: Michigan Nanofabrication Facility, University of MichiganNNIN REU Principal Investigator: Professor L. Jay Guo, Applied Physics and Electrical Engineering

& Computer Science, University of Michigan, Ann ArborNNIN REU Mentor: Brandon Lucas, Applied Physics, University of Michigan, Ann ArborContact: [email protected], [email protected]

Fabrication of Nanoparticle Arrays:All glass substrates were cleaned in a 1:1 piranha solution

of H2O2:H2SO4 for 15 minutes. Once soaked in solution, the substrate was rinsed with copious amounts of DI:H20 and finally blown dry using N2. A layer of mR-I 8020 photoresist was spin-coated onto the surface of the substrate to the appropriate thickness then baked on a hot plate at 140°C for 5 minutes to remove any solvents. A nanoimprinter was used to imprint a mold with periodic square nanoparticle features with in-plane widths of ~ 110 nm and particle spacing of ~ 100 nm directly onto the prepared substrate at 180°C and 670 psi. After imprinting, the residual layer of resist was removed using reactive ion etching (RIE). A select recipe of 20 sccm of O2 at 50 watts of power and 20 mTorr was used to etch the polymer away. Following the polymer etch, a very thin layer of titanium followed by a layer of metal was deposited onto the surface of the substrate using an electron beam evaporator. Metals generally have poor adhesion properties, therefore the titanium adhesion layer was essential in fabricating metallic nanoparticles directly onto a substrate. Lift-off was performed by placing the substrate in a beaker of acetone and then placing the beaker of acetone in an ultrasonic bath. The finished sample was rinsed with methanol and IPA, and dried with N2 to completion. SEM images of our fabricated nanoparticle arrays are illustrated in Figure 1.

Figure 1: Fabricated metallic nanoparticle arrays.


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

Testing:Optical extinction measurement tests were conducted using

a Nikon TE300 Eclipse inverted microscope (20x objective) with transmitted broadband light coupled into an Ocean Optics SD2000 fiber-coupled spectrometer. Adhesion layer testing was first conducted to study the effects on the resonance of the metallic nanoparticles. 20 nm gold nanoparticles, as well as silver nanoparticles with adhesion layers of 0 nm, 1 nm and 3 nm were tested. As shown in Figure 2, a consistent resonance peak shift to shorter wavelengths was observed as the thickness of the titanium layer increased. In addition, the 1 nm adhesion layer uniquely exhibited a stronger resonance than other adhesion layer thicknesses. The effect of the metal thickness on the LSPR was also studied. Gold and silver nanoparticles of ranging thicknesses using a 1 nm titanium adhesion layer were fabricated and tested. Figure 3 illustrates the consistent shift of the wavelength peak to shorter wavelengths as the metal

thickness increased. Silver nanoparticles exhibit a sharper resonance peak and therefore allow for better shift detection than gold nanoparticles. The consistent trend in shifts shows that NIL is able to produce metallic nanoparticle arrays that are very easily tunable.

We lastly performed refractive index testing to test the sensitivity of our nanoparticles. A flow cell setup was used to flow different concentrations of a glycerol/water solutions over our sensing platforms. The change in the refractive index of the surface resulted in a shift of the resonance peak. Silver nanoparticles exhibited higher sensitivity than gold nanoparticles, as can be seen in Figure 4. A greater shift in wavelength resulted in response to a change in refractive index for silver nanoparticles.

Results and Conclusions:During the duration of this project, we not only successfully

demonstrated the fabrication of nanoparticle arrays by nanoimprint lithography, but we additionally identified the optimal adhesion layer thickness for LSPR and examined the LSPR dependence on metal thickness of nanoparticles. And lastly, by performing refractive index sensing experiments, we were able to observe a stronger response from Ag nanoparticles of smaller height.

Acknowledgements:I would like to thank Professor L. Jay Guo, Brandon Lucas,

Dr. Sandrine Martin, Dr. Jin-Sung Kim, Myung-Gyu Kang, Guo Nanogroup, Michigan Nanofabrication Facility, the National Science Foundation and National Nanotechnology Infrastructure Network for all their help and support.

Figure 4: Refractive index testing.

Figure 2: Adhesion layer testing on 20 nm silver nanoparticles.

Figure 3: Metal layer testing on gold nanoparticles.


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract: Electron beam lithography was used to pattern a 1 x 1 mm matrix of holes with 300 nm diameters in silicon nitride. This master stamp was used in a nano-imprint process to cast PMMA resist dots on an exchange coupled composite magnetic bilayer, which when ion-milled, produces separated islands of magnetic material (100 nm wide with a 1 µm period). This nano-imprinting process allows inexpensive production of patterned media by repeatedly and quickly casting patterns that were slowly and painstakingly created. The developed nano-imprinting technology could be applied to any other product from logic devices to quantum computing which requires large numbers of inexpensive nano-scale features.

Introduction:

In the quest for 1 terabit per inch square magnetic hard drive data densities, patterned media offers the way forward. Because the grain size of the magnetic material is effectively limited to ~ 8 nm by the superparamagnetic effect, and in continuous media the signal to noise ratio depends on the number of grains making up each bit, the data density of continuous media is limited. Patterned media allows one grain per bit by physically separating grains from each other, meaning at least a 10 fold increase in data density.

The magnetic material itself was an exchange coupled composite (ECC). Magnetic materials with a high anisotropy constant offer long data life, but are called ‘hard’ because it takes an enormously strong field to initially write data. ‘Soft’ materials are easy to write, but data degrades quickly. ECC media couples a soft material to a hard material to gain the benefits of both.

The fabrication process was aimed at mass production. Inexpensive PMMA was used throughout the process instead of a more expensive negative resist. A stamp was made via costly and slow electron-beam lithography, but was then repeatedly used to pattern casts in a nano-lithography process. This process combines the precise features offered by electron-beam lithography with the high throughput of nano-lithography.

A Combined Electron-Beam and Nano-Imprint Lithography Technique for the Affordable Creation of Exchange Coupled Composite Patterned Media

David W. CoatsPhysics, Harvey Mudd College

NNIN REU Site: Minnesota Nanotechnology Cluster, University of Minnesota, Twin CitiesNNIN REU Principal Investigator: Prof. Jian-Ping Wang, Electrical Engineering, University of Minnesota, Twin CitiesNNIN REU Mentor: Hao Meng, Electrical Engineering, University of Minnesota, Twin Cities Contact: [email protected], [email protected]


A 110 ± 15 nm thick layer of silicon nitride was deposited on a wafer, then spin coated with PMMA. Electron-beam lithography created patterned holes in the PMMA. This pattern was transferred to the silicon nitride layer by reactive ion etching with CF4. After cleaning off the rest of the PMMA with O2 etching the stamp was finished.

The cast was made by sputtering magnetic material onto a wafer, then spin coating with PMMA. The PMMA was hardened by baking at 180ºC for four hours. The PMMA was hardened so that it would not flow away or stick to the stamp during nano-imprinting.

When the stamp was pressed into the cast at 250 psi the stamp’s patterned holes were converted to patterned pillars on the cast. After pressing, the stamp was cleaned for reuse in an ultrasonic bath for 10 minutes, then under O2 etching for one minute.

The cast was then ion milled for 17 minutes. The PMMA pillars served as a resist such that only the islands of magnetic material protected by them remained afterwards.


As shown in Figure 1, the smallest dots patterned on the stamp were 50 nm in diameter. These smallest dots did not transfer well during nano-lithography, so 300 nm in diameter dots with 500 nm periods were used instead.

The depth of the holes in the stamp remains unknown. The maximum depth given by the AFM was 50 nm, with larger holes having larger depths. Whether this depth is accurate is an open question. Because the diameters were so small, the sides of the AFM probe might have physically kept the actual tip from touching the hole’s bottom.

Figure 2 shows minimal degradation to the stamp after five pressings. Some of the holes are filled in with PMMA, and dark smudges of PMMA cling to the sidewalls of other holes. This could be remedied by further O2 reactive ion etching. The primary method of


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

damage to the stamp came from dust causing cracks to form in the silicon during the pressing process.

The final patterned media is shown in Figure 3. On the macroscale, pattern uniformity was low with many dots missing from their intended spots. This was due to an overly long ion milling time. A shorter time would yield more uniform pattern, but larger dots.

Within the pattern, dot placement lacks uniformity as shown in Figure 4. This results from differences in the physical strength of the material under the PMMA resist pillars. The PMMA pillars protected many grains. During the ion milling process the weaker grains were blasted away and mere chance determined which of the grains survive. Thus the remaining grains were not in neat rows.

Because the nano-lithography was not done under clean room conditions, dust contamination was a major problem. If a piece of dust got between the stamp and cast, the pattern would not transfer because the stamp and cast were not in contact. Worse, because 250 psi was concentrated on one small dust particle, it would crack the silicon of the stamp and cast, which often lead to shattering.

Future Work:

The density of dots in the patterned media will be increased. The nano-lithography press will be brought into the clean room. This will allow smaller PMMA resist dots to transfer during nano-lithography which will improve the placement uniformity within the pattern. The ion milling time will have to be adjusted to the dot diameter and period for improved macroscale pattern uniformity.

Acknowledgements:

Prof. Jianping Wang; Xiaofeng Yao; Hao Meng; Dr. Jianmin Bai for AFM and MFM work; Weikang Shen for sputtering the magnetic material; NSF and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for funding.

Figure 1: The smallest patterned features on the stamp were 50 nm across.

Figure 2: An SEM of a stamp used 5 times showing minimal degradation.

Figure 3: Total pattern uniformity is low.

Figure 4: MFM showing uniformity of dot placement within the pattern is low.


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:Although conventional photolithography has seen numerous applications within the microelectronics industry, its functional potential within nanotechnology has thus far been limited by its inability to produce pattern features effectively in the sub-micron range. However, through implementation of a novel technique in which microscale patterns are written onto the photoresist layers of a semiconducting wafer using conventional photolithography and then metals are angularly deposited through the pattern gaps, we have broken the conventional photolithography resolution limit. Through the course of research, pattern features less than 80 nm in size have effectively been produced.

Fabrication of Sub-100 nm Structures Using Conventional Photolithography

Paul De AndradeBiomedical Engineering, Georgia Institute of Technology

NNIN REU Site: Center for Nanoscale Systems, Harvard UniversityNNIN REU Principal Investigator: Dr. Erli Chen, Center for Nanoscale Systems, Harvard UniversityNNIN REU Mentor: Dr. Jiangdong Deng, Center for Nanoscale Systems, Harvard UniversityContact: [email protected], [email protected]

Materials & Process:

A detailed diagram of the fabrication is shown in Figure 1.

Spin Coating & Photolithography: Research was performed using silicon (Si) wafers with a layer of silicon dioxide (SiO2) on the surface. The wafers were spin-coated with a layer of lift-off resist (LOR 20B), a thicker resist with a high dissolution rate during development. They were then spin-coated with photoresist (PR 1805). Conventional photolithography was used to expose the samples through a photomask with 1 µm pattern lines on it, and the samples were subsequently developed in order to transfer the pattern into the layers of photoresist. Because of the higher dissolution rate of the LOR 20B, an undercut was produced beneath the top photoresist layer, as shown in Figure 1e.

Thermal Evaporation: Samples were mounted into the upper stage of the thermal evaporator (Sharon) for the metal deposition. A thin layer of chromium (Cr) was first deposited to enhance subsequent gold adhesion. Two gold sources were used to deposit metal into the resist lines from two opposing angles. The deposition angle was varied by changing the spacing of two gold sources. In this experiment, the distance of the two gold sources was within 8.2 ± 0.5 cm.

Resist Removal: The resist layers were removed from the surface of the samples. This was accomplished by heating two beakers of MicroChem Remover PG to within a temperature of 60-70ºC and immersing the samples multiple times.

Reactive Ion Etching: To remove the thin layer of chromium within the nano-gap, samples were immersed in a wet etchant for 7 seconds. The samples were inserted into a reactive ion etcher (RIE, NEXX Inc.), which etched down the wafer surface around the gold pattern lines. A wet etchant was again used in order to remove any gold from the surface, leaving only the pattern etched into the wafer surface.

Figure 1: a) Diagram of initial Si/SiO2 wafer, b) Coating with LOR 20B, c) Coating with PR 1805, d) Exposing with mask-aligner, e) Resist development, f) Gold deposition, g) Resist removal, h) Etching.


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

Results & Discussion:

Optimizing Photolithography: During the photolithography process, the most significant determinant of pattern transference accuracy from mask to photoresist was the mask alignment exposure time. For the silicon samples coated with LOR 20B and photoresist S1805, an exposure time of 2.4 seconds was found to transfer the pattern with the highest accuracy and consistency while preventing excessive dissolution of the lift-off resist during development.

Manipulating Gap Size: Upon extensive testing and analysis, both the surface photoresist gap size and metal source spacing within the thermal evaporator were determined to have significant impact upon the produced nano-gap. The thinnest (1 µm) pattern gaps consistently yielded the smallest nano-gaps between deposited metal strips. It was also found that the nano-gap width could be controlled by varying the spacing between the two metal sources within the thermal evaporator. In Figure 2, by implementing a source spacing of 8.4 cm, a nano-gap of approximately 78 nm was achieved and imaged using a scanning electron microscope (SEM).

The Final Product: Upon measurement, the width of the nano-gap was found to be 71.6 nm (Figure 3), approximately 4 nm thinner than when the gap was imaged prior to reactive ion etching. A possible cause of this variance was that because of the angular deposition of the gold, the line edges may have lain at an angle to the surface of the wafer, therefore tapering inwards (towards center of nano-gap) near the surface and further reducing the width of the etched area. The height of the gap was recorded at 52.2 nm, a value which for the purposes of this research was satisfactory but could potentially be increased by longer RIE etching.

Conclusions and Recommendations:

Through the application of this novel cross-metal deposition into a lift-off resist undercut, the consistently successful fabrication of a nano-gap of less than 80 nm was evaluated. It was also shown that simply varying the metal source spacing could lead to a significant change in gap size using this technique. Therefore, the pattern feature size is no longer limited by diffraction as it has previously been in conventional photolithography.

Upon further research and refinement of testing conditions, this technique has the potential to open new doors for conventional photolithography. Adjusting exposure time and metal source spacing allows the process to be customized for applications requiring nanoscale features of specific sizes. Through slight manipulation of the reactive ion etching recipe, the produced wafer may be delicately refined for actual semiconductor device implementation. Additionally, creating a poly-dimethylsiloxane (PDMS) stamp of features created using our technique would allow for fast and effective reproduction of patterned features.

Acknowledgments:

Dr. Jiangdong Deng; Dr. Erli Chen; Dr. Kathryn Hollar; National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

Figure 2: SEM: Nano-gap within LOR undercut showing a surface photoresist gap of 1.17 µm and a nano-gap of 77.9 nm.

Figure 3: SEM: Cross-sectional view nano-gap width = 71.6 nm, height = 52.2 nm.


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:A potential approach to higher storage density for magnetic recording is patterning the thin film magnetic medium into arrays of physically isolated islands, which can then be magnetized to represent bits. In this study, we produced prototype patterned media by using a focused ion beam (FIB) to cut arrays of islands into a CoCrPt-alloy perpendicular magnetic thin film. We then examined the physical properties and magnetic characteristics of the islands with atomic force microscopy (AFM) and magnetic force microscopy (MFM), respectively. By varying the milling time of the patterns, we were able to produce trenches with various depths. Results indicated that even trench depths as shallow as 2-3 nm were able to sustain proper magnetic isolation of islands. Attempts at patterning smaller islands demonstrated the necessity of shallow, narrow trenches for well-defined islands with smaller periods.

Introduction:

Magnetic recording, widely used for data storage in modern electronic devices, involves representing a bit of data by the alignment of a magnetic domain in a thin magnetic film. However, data density is limited because the bits must exceed a certain volume to avoid the superparamagnetic effect, in which random thermal energy eventually causes the magnetization to spontaneously switch direction. Patterned media, in which the bits are physically isolated, may be able to overcome this limitation because the increased stability of the domains allows smaller bit volume and therefore greater data density [1].

FIB is a patterning method that involves using a focused Ga+ ion beam to cut trenches in the media. The irradiated regions are unable to sustain magnetization in the absence of field, thus isolating the islands. Decreasing milling (exposure) time allows for narrower trenches, which are advantageous for patterning smaller islands; however, insufficient exposure may lead to shallow trenches and incomplete island isolation [2].

In this work, we varied milling time of the patterns

Nanoscale Focused Ion Beam Patterning and Characterization of Perpendicular Magnetic Recording Media

Irene HuElectrical Engineering, Princeton University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford UniversityNNIN REU Principal Investigator: Robert Sinclair, Materials Science and Engineering, Stanford UniversityNNIN REU Mentor: Unoh Kwon, Materials Science and Engineering, Stanford UniversityContact: [email protected], [email protected]

to produce trenches of various depths and widths, and analysed the resulting island magnetizations to determine the trench depth necessary for magnetic isolation. We also produced patterns with islands of various sizes to determine the trench parameters necessary for small, well-defined islands. Doing so will help us develop optimal parameters for FIB patterning of thin magnetic film.

Procedure:

Using the FIB to accelerate a 1 pA Ga+ current through a 30 keV voltage, 10 x 10 uniform square grids were patterned on a 20 nm thick CoCrPt-alloy thin film. Patterns with a period of 500 nm were milled for 43.5 s, 2 min 32 s, and 4 min 21 s. Milling time scaled to size, allowing equivalent dosage, was applied to produce patterns with period of 300, 230, 180, 150, 120, and 100 nm. The sample was then dc-magnetized, and physical parameters and magnetic characteristics were confirmed with AFM and MFM, respectively.


Trench width and depth varied with the patterns, but were on the order of 50-60 nm and 2-3 nm width and depth, respectively, for the lowest dosage, 60-80 nm and 3-4 nm for the middle, and 100 nm and 4-5 nm for the highest. Previous studies indicated that trench depths of around 4-5 nm were required for magnetic isolation of islands [2]. However, as apparent from the cross section of a 230 nm period pattern in Figure 1 and the corresponding MFM scan in Figure 2, even depths of around 2-3 nm appear to be sufficient for this media. In fact, MFM data demonstrated proper magnetic isolation for almost all patterns.

It has been hypothesized that the vanishing of magnetic properties in trenches, leading to magnetic isolation of islands, results not only from media removal, but also from disruption from collisions and ion implantation [2]. Thus, even though the media is 20 nm thick, these shallow trenches are already sufficient for magnetic isolation.

For the smaller patterns (p < = 180 nm), a higher


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

dosage caused the trench depth to decrease, possibly due to the overlapping of trenches shaving off the tops of islands. This can be seen in Figure 3, which compares the cross sections of 100 nm period patterns with different dosages. Nevertheless, even with this loss of definition, patterns with longer milling time still showed some magnetic isolation, as shown in Figure 4.

Conclusion:

Dosage producing trenches as shallow as 2-3 nm appears to be sufficient to magnetically isolate islands. With lower dosage, trenches are also narrower, allowing patterns with smaller periods. However, reducing the period further requires reducing trench width even more to prevent loss of physical definition. Thus, to achieve good patterns with smaller periods, increasingly low dosages are required.

The smallest pattern produced here, with p = 100 nm, corresponds to a data storage density of 64.5 Gb/in2. A much smaller period would be required for patterned media to be competitive with current technology; p = 50.5 nm is required for 230 Gb/in2, the expected data density of perpendicular magnetic recording by 2007. However, as already stated, there are substantial difficulties in scaling down period. Further studies involving even less dosage are necessary to determine the absolute

minimum for magnetic isolation. This then determines the minimum trench width, and therefore the smallest period, achievable at the current ion beam size.

Acknowledgements:

Unoh Kwon; Professor Sinclair; the Sinclair group; Joon Seok Park; Max Gage; Mike Deal; the REU staff; National Nanotechnology Infrastructure Network (supported by NSF).

References:[1] M. Albrecht et al. “Thermal stability and recording properties

of sub-100 nm patterned CoCrPt perpendicular media.” Journal of Applied Physics 91.10 (2002): 6845-6847.

[2] C. T. Rettner et al. “Magnetic Characterization and Recording Properties of Patterned Co70Cr18Pt12 Perpendicular Media.” IEEE Trans. on Magnetics 38.4 (2002): 1725-1730.

Figure 1: AFM cross section of 230 nm period pattern at lowest dosage.

Figure 2: AFM/MFM image of 230 nm period pattern at lowest dosage.

Figure 3: AFM cross sections of 100 nm period patterns with low and high dosage.

Figure 4: AFM/MFM image of 100 nm period pattern at highest dosage.


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:Experimentation of desorption-ionization mass spectrometry was tested on nanoporous polymer. Matrix assisted laser desorption-ionization (MALDI) is used in a wide series of applications having the most use in specific biological experiments such as peptide sequencing and the analysis of small biomolecules. The goal of this experiment was to improve the accuracy of analyzing larger biomolecules as well as improve the sensitivity typically achieved with other matrices. In this experiment nanoporous polymer was substituted as an ultraviolet light absorbing matrix. In MALDI the matrix used typically absorbs in the frequency of the laser being used, and indirectly transfers the energy of the laser to the sample, causing the sample to sublimate and ionize. In our experiment the MALDI micro MX with a nitrogen laser (337 nm) was used to ionize each sample plated on our nanoporous polymer. The recipe for our nanoporous polymer was slightly modified throughout the experimentation in order to improve in its operation as a substituted matrix.

Introduction:

Mass spectrometry is a vital tool in modern biology. A specific form of mass spectrometry known as MALDI (matrix assisted laser desorption-ionization) is used today to analyze proteins, peptides, polysaccharides, nucleic acids, synthetic polymers, bacteria, and other small molecules. In MALDI, sample preparation involves mixing the organic molecule of your choice with a solvent, ie. acentonitrile or ethanol. The sample is then mixed with an ultraviolet absorbing compound, or UV matrix. After being placed in a vacuum chamber, the solvent evaporates leaving only the substrate and the matrix. The substrate is then shot with a pulse laser and the UV absorbing matrix absorbs the energy from the laser and transfers it to the analyte, causing it to sublimate and ionize. After ionization the analyte is placed in an electric field and accelerated towards the detector. By calculating the speed of the analyte and magnitude of the electric field, the mass/charge ratio can be determined.

Desorption-Ionization Mass Spectrometry on Nanoporous Polymer

John KrogerBiomedical Engineering, Georgia Institute of Technology

NNIN REU Site: Penn State Nanofabrication Facility, The Pennsylvania State UniversityNNIN REU Principal Investigator: Tony Jun Huang, Engineering Science and Mechanics, The Pennsylvania State UniversityNNIN REU Mentor: Vincent Hsiao, Engineering Science and Mechanics, The Pennsylvania State UniversityContact: [email protected], [email protected]

Currently there is much research in improving MALDI. One problem with MALDI is that there is a limit on the size of the molecule that can be accurately measured. Large biomolecules such as proteins, as well as small molecules such as monomers are difficult to measure with MALDI and other methods are typically used. While the use of a matrix is vital in MALDI, the added substrate causes background noise that interferes greatly in the lower mass region, and causes inaccuracies at the opposite end. The use of nanostructure particles as a substituted matrix has been seen with structures such as nanoporous Si, oxidized carbon nanotubes, nanostrucured-column void Si, and gold nanoparticles. With the success of these other nanostructures in mind, we developed a nanoporous polymer and tested its effectiveness as a substituted matrix. Through our nanoporous polymer we hope to facilitate future sample preparation, analyze larger and smaller biomolecules, and obtain more sensitive and accurate sample readings.


Our nanoporous polymer was created by mixing the following ingredients: anthracenetriol (polmer dye), a-cyano-4-hydroxycinnamic acid (peptide dye), vinyl-2-pyrrolidone, rose bengal, n-phenylglycine, liquid crystal,

Figure 1


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

acetone, sitane, and dipentaerythritor penta/hexa-acrylate (molymer). The polymer mixture was then plated onto a glass substrate and exposed to an argon laser for one minute; the setup is shown in Figure 1. The porosity of the polymer was determined by the angle of the beam splitter. After being exposed to the laser the polymer was cured under normal light for 24 hours. Our peptide was then plated onto our polymer substrate and analyzed with the MALDI micro MX.

Results and Future Work:

The results of our nanoporous polymer fabrication can be seen in Figures 2 and 3. No significant data was collected from MALDI experimentation; the only data points achieved were noise as seen in Figure 4. In the future we hope to construct a matrix substrate so that each organic material can be localized in a certain position. This would allow us to concentrate our organic material in a specific area for MALDI testing. This would not only facilitate localizing our laser pulses, but would also give us a more concentrated biospecies sample.

Conclusion:

The data collected did not show enough significance to support the role of nanoporous polymer as a substituted matrix in MALDI. However, this does not mean that nanoporous polymer will not work as a substituted matrix. In future work we would like to etch µL size wells into our glass substrate. This would prevent our peptide solution from dispersing and maintain a more concentrated solution in a smaller area, allowing the laser to better ionize our analyte. This, along with a better selection of peptides to perform initial testing, would be a great place to start in the future.

Acknowledgments:

I would like to thank Dr. Tony Jun Huang, Dr. Vincent Hsiao, Jinjie Shi, and Heike Betz for all their help throughout this project. I would like to also thank Penn State University and National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program for housing, funding, and most importantly the opportunity to be a part of this research experience.

References:[1] Kalkan, A; “Biomedical/analytical applications of deposited

nanostructured Si films”; Nanotechnology, 16, 1383-1391 (2005).

[2] Sudhir, P; “Identification of Peptides Using Gold Nanoparticle-Assisted Single-Drop Microextraction”; Anal. Chem, 77, 7380-7385 (2005).

[3] Ren, S; “Oxidized carbon nanotubes as matrix for matrix-assisted laser desorption/ionization time-of-flight mass spectrometric analysis of biomolecules”; Rapid Comm. In Mass Spectrometry, 19, 255-260 (2005).

[4] Wei, J; “Desorption-ionization mass spectrometry on porous silicon”; Nature, 399, 243-264 (1999).

Figure 2

Figure 3

Figure 4


PRO

CES

S &

CH

AR

AC

TER

IZAT

ION

Abstract:Electron beam lithography, with its incredible accuracy and patterning precision, is on the forefront of nanotechnology and nano-gap fabrication. Nano-gaps hold promise for a variety of reasons. In molecular detection, smaller gaps can detect fewer particles, having applications in airport security, chemical warfare and environmental monitoring. Nano-gaps could also be used for better DNA detection and research in molecular electronics. To achieve these applications, several methods have been undertaken to help increase the resolution of electron beam lithography systems, including the use of thin resist thicknesses, beam interference, limiting apertures and thin membranes.

In this project, the state-of-the-art JEOL JBX-9300FS 100kV system at Georgia Tech was used to obtain high resolution and very precise patterns. A drawback of this system is the backscattered electrons, which can expose unpatterned resist, blurring patterns. This drawback can be nearly eliminated by using thin membranes. To obtain these membranes, the wafer was selectively etched away to leave a layer of silicon nitride with a thickness of a few hundred nanometers. The e-beam pattern was then written on the membranes and the sample underwent a gold lift-off process. Characterization, using the scanning electron microscope (SEM) and atomic force microscope (AFM), has shown successful fabrication of nanometer scale gaps.

Procedure:

The first step in the process was to fabricate the membranes for writing. In order to do this, a process for efficiently and repeatedly achieving these membranes had to be finalized. For our process, type <100> wafers were used due to their etching properties, and it was found that double side polished wafers were also needed.

Our membranes were made of silicon nitride, Si3N4. The Si3N4 was deposited on both sides of the wafer. The Si3N4 served a dual purpose. On the front side of the wafer, the nitride would become the membrane

Fabrication of Nanometer-Scale Gaps on Thin Nitride Membranes using Electron Beam Lithography

David M. SchlunekerMechanical/Electrical Engineering, Rose-Hulman Institute of Technology

NNIN REU Site: Microelectronics Research Center, Georgia Institute of TechnologyNNIN REU Principal Investigator: Dr. Raghunath Murali, Microelectronics Research Center, Georgia Institute of Technology NNIN REU Mentor: Farhana Zaman, Microelectronics Research Center, Georgia Institute of Technology Contact: [email protected], [email protected]

after etching. On the back side, the Si3N4 worked as an etch mask. The front layer of nitride was very thin; experimentation was performed with thicknesses from half of a micron to as thin as a few hundred nanometers. The back layer was coated with approximately 1 µm of nitride. For our trials, the deposition was performed with a plasma enhanced chemical vapor deposition system (PECVD).

After deposition, standard photolithography was performed on the front of the wafer. Our pattern was then developed and removed, leaving the pattern in exposed silicon with the developed resist as an etch mask for the nitride etch. At first, a reactive ion etcher (RIE) was used to etch through the nitride, but it could not be kept clean enough to give a good etch. Using an inductively coupled plasma etcher (ICP), the nitride etch worked much better, giving clear pattern transfer and etching all the way through the nitride, exposing the silicon substrate.

With the nitride removed, the silicon etch could be performed. This process used a wet etch with 30% potassium hydroxide (KOH). Due to the <100> orientation, the exposed silicon was etched at a 54.74° angle, as the profile in Figure 1 shows. Etch rates of 1 to 1.6 µm per minute were obtained with etch selectivity of nearly 1000:1, silicon to silicon nitride. After the

Figure 1: This profile shows the 54.7° etch angle through the wafer.


PR

OC

ESS &

CH

AR

AC

TERIZATIO

N

nitride etch was completed, the membranes were ready for use.

The next step was to use the JEOL JBX-9300FS electron beam lithography tool to write our pattern onto the membrane and substrate. For our pattern, we put part on the membrane and part on the substrate in order to see the contrast between the membrane and substrate. With the exposure and development completed, the imaging could be done. Imaging was done using an SEM. The AFM could not give us the resolution needed.

on the membrane came out very clear. The membranes produced smaller gaps with much greater precision.

Future Work:

With the promising results and possibilities of this project, we would like to see more work done to find a better etching process. Another idea is to use a metal etch mask for better selectivity. Finally, the use of sacrificial polymers in construction of these membranes is another area of interest.

Acknowledgements:

I would like to acknowledge and thank Dr. Raghunath Murali, Farhana Zaman, Cheng-Tsung Lee, Dr. James Meindl, Ms. Jennifer Tatham Root, and the clean room staff for their guidance, help and time during this project. Finally, thank you to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program.

Figure 2: The smallest gap achieved, 5.78 nm, is shown here.


From our experimentation, very promising results were obtained. The process for fabrication of the membranes was established and optimized. The membrane squares could be fabricated with a high yield per wafer, efficiently, and fairly quickly. The process worked very well and success was achieved for every trial using the double side polished wafers.

The results from patterning were also very promising. While we were able to successfully fabricate gaps as low as 6 nm, there is good reason to believe narrower gaps can be achieved. Figure 2 shows the smallest gap achieved in this series of trials, less than 6 nm. By varying the dosage applied, the gap size varied until a point was reached when the dosage was too high and the pattern merged. While the gap sizes we achieved were good, the most compelling results from our research can be seen in Figure 3 and 4. These SEM images were taken right at the border of where the membrane and substrate-backed patterns met. A clear distinction can be seen between the patterns made on the substrate-backed nitride and the membranes. The patterns that were written on the substrate were merged at a much lower dose, possibly due to the backscatter. The patterns were also skewed and tilted. On the other hand, the patterns that were written

Figure 3: The quality of the substrate written patterns is lower then the membrane patterns.

Figure 4: This image shows the patterns written on the membranes.

National Nanotechnology Infrastructure Network Research Experience for Undergraduates 2006 Research Accomplishments page 128

The NNIN REU pictures on Cornell Campus were taken

by Dede Hatch, Photographer

Photographs at other sites were taken by site staff.

2006 NNIN REU Program Thanks for a great summer!

page 129 National Nanotechnology Infrastructure Network Research Experience for Undergraduates 2006 Research Accomplishments

Index(NNIN REU Interns are in Bold)

•NNIN Sites•Cornell Univ. .. ..10, 16, 26, 30, 38, 40, 52, 80, 82, 84Georgia Tech .. .. .. .. .. .. .. .. .. .. . 28, 32, 46, 72, 126Harvard Univ. . .. .. .. .. .. .. .. .. .. .. 50, 106, 114, 120Howard Univ. . .. .. .. .. .. .. .. .. .. .. .. .. 42, 48, 62, 94Pennsylvania State Univ. .. .. .. .. . 22, 24, 44, 60, 124Stanford Univ. .. .. .. .. .. .. .. .. 18, 56, 102, 104, 122UCSB . .. .. .. .. .. .. .. .. .. .. .. 12, 76, 78, 86, 108, 112Univ. of Michigan .. .. .. .. .. .. .. . 34, 54, 64, 70, 116Univ. of Minnesota .. .. .. .. .. .. .. .. .. . 2, 4, 8, 74, 118Univ. of New Mexico . .. .. .. .. .. 58, 68, 88, 100, 110Univ. of Texas. .. .. .. .. .. .. .. .. .. .. .. .. .. 6, 66, 92, 98Univ. of Washington .. .. .. .. .. .. .. 14, 20, 36, 90, 96

•A•Adams, Michael . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 114

Agah, Ali .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 18Ainyette, Alina. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 48

Awschalom, David .. .. .. .. .. .. .. .. .. .. .. .. .. 86Aydil, Eray .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 74Aziz, Michael .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 50

•B•Badaire, Stephane . .. .. .. .. .. .. .. .. .. .. .. .. .. 52Bajwa, Ravneet .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 30

Bakir, Muhannad .. .. .. .. .. .. .. .. .. .. .. .. .. .. 46Ballinger, Andrew .. .. .. .. .. .. .. .. .. .. .. .. .. 32

Barocas, Victor .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 4Bayer, Carolyn .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 98Bazille, Lawrence . .. .. .. .. .. .. .. .. .. .. .. .. . 94

Becker, Udo .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 54Bell, David .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 114

Berezovksy, Jesse .. .. .. .. .. .. .. .. .. .. .. .. .. .. 86Berlin, Jana .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 58Bischof, John .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 8Blosser, Matthew. .. .. .. .. .. .. .. .. .. .. .. .. . 50

Boercker, Janice . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 74Boettcher, Shannon .. .. .. .. .. .. .. .. .. .. .. .. .. 78Böhringer, Karl F. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 96Bonthera, McIntosh .. .. .. .. .. .. .. .. .. .. .. .. 52

Bowers, John E. . .. .. .. .. .. .. .. .. .. .. .. .. .. . 112Brearley, Adrian J.. .. .. .. .. .. .. .. .. .. .. .. .. .. 58Breitbach, Anthony S.. .. .. .. .. .. .. .. .. .. . 54

Brown, Devin .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 32Bryan, Sarah .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 56

•C•Catchmark, Jeffrey .. .. .. .. .. .. .. .. .. .. .. .. .. 44Chambers, Matthew . .. .. .. .. .. .. .. .. .. .. .. . 108Chen, Erli .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 120Cheng, An . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 24Cheng, Wenlong. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 26Chin, Christine . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 116

Clarke, David .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 108Coats, David W. . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 118

Cohen, Itai. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 80Cohn, Alicia. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 58

Cordovez, Bernardo . .. .. .. .. .. .. .. .. .. .. .. .. 10Cornell, Eva .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 2

Corso, Christopher .. .. .. .. .. .. .. .. .. .. .. .. .. 28Cortes-Jimenez, Sonia Y. .. .. .. .. .. .. .. .. . 60

Costello, Kelly .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 4

Cramer, George .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 34

Crozier, Ken . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 106


•D•D’Silva, Joseph .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 6

Dai, Hongjie . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 104Daise III, Henry .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 62

Dang, Ying Yi .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 64

Datye, Abhaya. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 68Daugherty, Patrick S. .. .. .. .. .. .. .. .. .. .. .. .. 12Dauskardt, Reinhold H. .. .. .. .. .. .. .. .. .. .. .. 56De Andrade, Paul . .. .. .. .. .. .. .. .. .. .. .. .. 120

Demirel, Melik .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 22Deng, Jiangdong .. .. .. .. .. .. .. .. .. .. .. .. .. . 120Deng, Luxue Rose. .. .. .. .. .. .. .. .. .. .. .. .. 100

Desai, Amit .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 60

•E•Eid, Khalid .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 44El-Sayed, Mostafa A. .. .. .. .. .. .. .. .. .. .. .. .. 72Erickson, David .. .. .. .. .. .. .. .. .. .. .. .. .. 10, 16

•F•Fang, Alexander . .. .. .. .. .. .. .. .. .. .. .. .. .. . 112Fox, Michael J.. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 36

Fung, Wayne Y. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 64

•G•Gachelet, Eliora .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 14Gage, David Maxwell . .. .. .. .. .. .. .. .. .. .. .. 56Gerbode, Sharon .. .. .. .. .. .. .. .. .. .. .. .. .. .. 80Gokirmak, Ali . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 38Goldman, Rachel S. . .. .. .. .. .. .. .. .. .. .. .. .. 70Griffin, James .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 42Griffin, Peter .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 18Guo, L. Jay .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 116

•H•Hagemeier, Jenna . .. .. .. .. .. .. .. .. .. .. .. .. 102

Han, Sangwook .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 92Haque, Aman .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 60Harjanto, Dewi .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 8

Harris, Gary L. .. .. .. .. .. .. .. .. .. 42, 48, 62, 94Haynes, Christy .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 2Heitsch, Andrew. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 66Henry, Nathan . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 38

Hernandez, Sarah C. . .. .. .. .. .. .. .. .. .. .. . 66

Hou, Alex Tou-Hung .. .. .. .. .. .. .. .. .. .. .. .. 30Hsiao, Vincent . .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 124Hsu, Chih-Peng .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 36Hu, Irene .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 122

Huang, Tony Jun .. .. .. .. .. .. .. .. .. .. .. .. .. . 124Huffaker, Diana .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 110Hunt, William D. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 28

•J•Jewell-Larsen, Nels .. .. .. .. .. .. .. .. .. .. .. .. .. 36Jin, Yu .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 70Jung, Joo-Yun .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 92

•K•Kabadi, Suraj .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 10

Kamanzi, Albert . .. .. .. .. .. .. .. .. .. .. .. .. .. . 40

Kan, Edwin .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 30Karim, Ayman . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 68Kaviani, Athra . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 12

Kidd, Brian .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 20King Jr., Calvin R.. .. .. .. .. .. .. .. .. .. .. .. .. .. 46Kirschke, Elaine . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 14

Korgel, Brian A. . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 66Kroger, John . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 124

Kurdak, Cagliyan .. .. .. .. .. .. .. .. .. .. .. .. .. .. 70Kwon, Unoh. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 122

page 131 National Nanotechnology Infrastructure Network Research Experience for Undergraduates 2006 Research Accomplishments

•L•Laghumavarapu, Ramesh .. .. .. .. .. .. .. .. .. . 110Lannon, Herbert. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 16

Lavenson, David .. .. .. .. .. .. .. .. .. .. .. .. .. . 68

Lawrence, Juliet. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 42

Lee, Jaegoo .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 30Lipson, Michal .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 40Lu, Christina .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 18

Lu, Wei .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 64Lu, Xianmao .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 90Lu, Yuerui .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 104Lucas, Brandon .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 116Luo, Dan .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 26

•M•Mach, Albert . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 20

Malloy, Kevin .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 88Mamishev, Alexander .. .. .. .. .. .. .. .. .. .. .. .. 36Mangan, Niall M. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 70

Marquis, Bryce .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 2Martin, Myranda .. .. .. .. .. .. .. .. .. .. .. .. .. . 44

Martin, Robert Patrick .. .. .. .. .. .. .. .. .. .. 22

Martin, Ryan .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 88Mathew, Esha .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 24

Matthews, Jason. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 48McVittie, James P.. .. .. .. .. .. .. .. .. .. .. .. .. . 102Mehta, Manish .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 110Meng, Hao. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 118Moon, Hongdae .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 6Moskovits, Martin. .. .. .. .. .. .. .. .. .. .. .. .. .. 76Murali, Raghunath .. .. .. .. .. .. .. .. .. .. .. .. . 126Murphy, Katrina M. .. .. .. .. .. .. .. .. .. .. .. .. 72

•N•Nakamura, Yasuhide .. .. .. .. .. .. .. .. .. .. .. .. 74Neikirk, Dean P. . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 92Nishi, Yoshio .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 102Norvell, Emily .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 76

•O•Osinski, Marek .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 100Ouzounova, Totka . .. .. .. .. .. .. .. .. .. .. .. .. .. 82

•P•Peppas, Nicholas A. . .. .. .. .. .. .. .. .. .. .. .. .. 98Petrina, Stephanie .. .. .. .. .. .. .. .. .. .. .. .. .. 78

Phillips, Jamie . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 34Plummer, James . .. .. .. .. .. .. .. .. .. .. .. .. .. .. 18Poitras, Carl .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 40Pratt, Erica. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 80

•R•Reason, Matthew .. .. .. .. .. .. .. .. .. .. .. .. .. .. 70Renock, Devon .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 54Ressler, Alice .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 2Rice, Jeffery J.. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 12Roberson, Leila Joy .. .. .. .. .. .. .. .. .. .. .. .. 82

•S•Sanda, Hiroyuki . .. .. .. .. .. .. .. .. .. .. .. .. .. . 102Sanders, Anthony .. .. .. .. .. .. .. .. .. .. .. .. 104

Schierhorn, Martin .. .. .. .. .. .. .. .. .. .. .. .. .. 76


Schluneker, David M. .. .. .. .. .. .. .. .. .. .. 126

Shastry, Ashutosh .. .. .. .. .. .. .. .. .. .. .. .. .. .. 96Shodeinde, Tajudeen . .. .. .. .. .. .. .. .. .. .. . 46

Siddiqui, Jeffrey J. .. .. .. .. .. .. .. .. .. .. .. .. .. 34Sinclair, Robert .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 122Smith, Cary .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 84

Smolyakov, Gennady .. .. .. .. .. .. .. .. .. .. .. . 100So, Eric .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 22Stern, Julie .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 106

Stoafer, Chris .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 108

Stroock, Abraham . .. .. .. .. .. .. .. .. .. .. .. .. .. 52Stucky, Galen D. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 78Sugiyama, Masano .. .. .. .. .. .. .. .. .. .. .. .. .. .. 4Swaim, Jon. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 26

•T•Tabor, Christopher E. .. .. .. .. .. .. .. .. .. .. .. .. 72Taylor, Crawford .. .. .. .. .. .. .. .. .. .. .. .. .. .. 62Taylor, Dane .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 96

Thomas, Wendy .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 20Tiwari, Sandip . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 84Toyli, David. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 86

Traxler, Beth. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 14Turner, Brendan. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 110

•U•Umbach, Christopher (Kit) .. .. .. .. .. .. .. .. .. 82

•V•VanDelden, Jay S. . .. .. .. .. .. .. .. .. .. .. .. .. .. 84Vasudev, Amit . .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 98

Velten, Josef .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 88

Visaria, Rachana. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 8

•W•Wang, Jason. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 112

Wang, Jian-Ping . .. .. .. .. .. .. .. .. .. .. .. .. .. . 118Wang, Xinran .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 104Wu, Claude .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 28

•X•Xia, Younan .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 90Xu, Jian .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 24

•Y•Yakovenko, Olga .. .. .. .. .. .. .. .. .. .. .. .. .. .. 20Yang, Allen .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 16Young, Kaylie .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 92

•Z•Zaman, Farhana .. .. .. .. .. .. .. .. .. .. .. .. .. .. . 126Zhang, John X.J. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 6

The 2006 NNIN REU Research Accomplishments is also available on the

web in PDF, at: http://www.nnin.org/nnin_2006reu.html

This publication was formatted by Ms. Melanie-Claire Mallison, NNIN REU Program Assistant. She welcomes your comments at: [email protected]

Documents

National Nanotechnology Infrastructure Network · 2019-12-19 · page i National Nanotechnology Infrastructure Network Research Experience for Undergraduates, 2006 Research Accomplishments