Study of Current Measurment in Magnetic Force Microscopy by Margaret Hsin-Yi Wang Submitted to the Department of Electrical Engineering and Computer Science in partial fulfillment of the requirements for the degree of Master of Engineering in Electrical Engineering and Computer Science at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY February 2001 @ Margaret Hsin-Yi Wang, MMI. All rights reserved. The author hereby grants to MIT permission to reproduce and distribute publicly paper and electronic copies of this thesis document in whole or in part. A uthor ................ C ertified by ............ . . . . . . . . . . . . . . . . . .Department of Electrical Engineerd aRd ComputerScipCnce September'14, 2001 .................................... Rajeev J. Ram Associate Professor Thesis Supervisor Accepted by .................. Arthur C. Smith Chairman, Department Committee on Graduate Students MASSACHUJ JIl2TUIE OF TECHNOLOGY JUL 3 1 '1V LIBRARIES BARKER 3 Study of Current Measurment in Magnetic Force Microscopy by Margaret Hsin-Yi Wang Submitted to the Department of Electrical Engineering and Computer Science on September 14, 2001, in partial fulfillment of the requirements for the degree of Master of Engineering in Electrical Engineering and Computer Science Abstract Current measurement in magnetic force microscopy (MFM) senses the deflection of a magnetized scanning probe tip caused by the magnetic field from the current flow. Due to the high spatial resolution of MFM, accurate measurement of current can serve as a powerful tool for failure detection on small integrated circuits. This thesis presents theoretical analysis and experimental results on DC and AC current imaging. The movement of the scanning probe and the magnetic tip-sample interaction were studied and applied to the theory. Sensitivity for the two current imaging methods were determined to be 1 mA and 15 pA respectively. A general theory for non-linear tip-sample interaction was developed and improvement to current sensitivity through parametric gain was proposed and modeled. Thesis Supervisor: Rajeev J. Ram Title: Associate Professor 4 5 Acknowledgments There are many people without whom this thesis would not have been possible. First and foremost, I would like to thank my professor, Rajeev Ram, for his guidance these last four years. His zeal for learning is contagious. And even more than the abundance of knowledge he possesses, he is a man of much wisdom. I am very fortunate to have had the opportunity to work with such an intelligent, supportive, and fun-loving research group. I thank Mathew for being my partner-in-crime. The late nights in the AFM lab would not have been the same without him. I appreciate his attitude and outlook on life. Harry, a man with much technical expertise, has been truly selfless with his time. His openness and sincerity are refreshing. I am grateful for the walks to the Student Center with Kevin. He has shown me the meaning of carpe diem with his life. Peter, my fellow desk partner, has been most patient these last few months with my taking over our desk. His ability to balance graduate life with other deserving priorities has been inspiring. And for the rest of the group, Farhan, George, Tom, and Seung-Ho, I appreciate their friendships. I hope that their graduate experiences are most-fulfilling. So long guys. I would also like to thank the National Science Foundation for their support in tuition, equipment, and fees for using the AFM. My family and friends have always stood by my side and selflessly given me their love and encouragement. Papa Wang and Mama Wang have been my role models and I cannot ask more. Phil and Tif are very dear to me and I look forward to spending more time at home. Stan has been a loyal friend for the past 10 years. Jarter has been a supportive friend, through many laughters and good times. Frank has been a dear friend from afar. Bonnie and Ellen have seen the most of my ups and downs this year and I appreciate their patience and care for me. I am truly blessed with wonderful people whom I cherish from the bottom of my heart. 6 Contents 1 1.1 1.2 Magnetic Force Microscopy 18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 . . . . . . . . . 19 1.1.3 M FM Tip . . . . . . . . . . . . . . . . . . . . . 20 1.1.4 Photodetection System . . . . . . . . . . . . . 20 1.1.5 Comparison to Other Magnetic Imaging Techniques . . . . 21 . . . . . . . . . . . . . . . . . . . . 22 DC Current Imaging . . . . . . . . . . . . . . . 25 . . . . . . . . . . . . . . . . . . 27 . . . . . . . . . . . . . . . . . . . . . 27 1.1.1 Overview 1.1.2 Movement of Cantilever Beam Point-Mass Model 1.2.1 2 17 Introduction 1.3 Flexural Beam Model 1.4 Thesis Overview 29 DC Current Imaging 2.1 Cantilever Dynamics . . . . . . . . . . . . . . . 29 2.1.1 Transverse Beam Equation . . . . . . . 30 2.1.2 Free Cantilever without Damping . . . . 34 2.1.3 Free Cantilever with Damping . . . . . 36 2.1.4 Comparison to the Point-Mass Model . 37 2.2 Magnetic Tip-Sample Interaction . . . . . . . . .. . . . .. . . . . 38 2.3 T heory . . . . . . . . . . . . . . . . . . . . . . . . .. . . .. . . . . 40 2.3.1 2.4 Comparison to the Point-Mass Model at the First Vibration Mode 41 Experim ent . . . . . . . . . . . . . . . . . . . . . . .. . . . .. . . 43 . . . . . . . . . . . . . . . . . . . . . . .. . . . .. . 43 2.4.1 Setup 7 8 CONTENTS 2.5 3 2.4.3 Comparison to Theory at the First Vibration Mode 2.4.4 Second Vibration Mode with Magnetic Tape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 . . . . . . . . . . . . . . 51 . . . . . . . . . . . . . . . . . . 51 Summ ary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 57 3.1 T heory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.1.1 Magnetic Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.1.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.1.3 Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Experim ent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 R esults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.3.1 Decomposition Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.3.2 Decomposed Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.3.3 Response to Varying Lift Heights . . . . . . . . . . . . . . . . . . . . . . . . . 66 Comparison to Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.4.1 M FM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.4.2 E FM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Summ ary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.3 3.4 3.5 Setup Method for Non-Linear Tip-Sample Interaction 83 4.1 T heory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.1.1 Limit to DC and AC Current Imaging . . . . . . . . . . . . . . . . . . . . . . 89 Application to Parametric Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.2.1 Principles of Parametric Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 4.2.2 Degenerate: wc=..p 92 4.2.3 Nondegenerate: w, 4.2 5 R esults AC Current Imaging 3.2 4 2.4.2 ............... . ............................. 93 4.3 Non-Linearity in AFM System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.4 Sum m ary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Conclusion 99 5.1 99 Summary CONTENTS 5.2 5.3 9 Directions for Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 5.2.1 M odeling 5.2.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Final W ords . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . 102 105 A Experimental Setup A.1 Instrument Setup ......... ...................................... 105 A.1.1 Atomic Force Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 A.1.2 Lock-in Amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 A.2 Sample Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 A.3 Data Acquisition and Processing Software . . . . . . . . . . . . . . . . . . . . . . . . 107 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 A.3.1 LabVIEW Code A.3.2 Matlab Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 A.4 Recipe for DC and AC Current Imaging . . . . . . . . . . . . . . . . . . . . . . . . . 107 B Magnetic Field Over a Wire 109 C Matlab Code 115 C.1 DC Current Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 C.2 AC Current Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 C.1.1 DCcurrent.m C.2.1 ACcurrent.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 C .2.2 decay.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 C .2.3 decayuse.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 C.3 Non-linear Current Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 C.3.1 degenerate.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 C.3.2 nondegenerate.m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 10 CONTENTS List of Figures 1-1 MFM setup for measuring current. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1-2 The first four vibration modes of a freely vibratingcantilever . . . . . . . . . . . . . . 20 1-3 Schematic drawing of a MFM tip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1-4 Schematic drawings of the position detection and the slope detection systems. 23 1-5 Comparison of magnetic imaging techniques . . . . . . . . . . . . . . . . . . . . . . . 24 1-6 The presence of a magnetic field causes a shift in the resonant f1requency, Awo = O- wo, affecting both the amplitude and phase responses..... . . . . . . . . . . 25 1-7 Illustration of force on an electric dipole. . . . . . . . . . . . . . . . . . . . . . . . . . 26 2-1 Bending moment exerted on a small segment of elastic beam. . . . . . . . . . . 30 2-2 Moment expressed as M = fA zodA. . . . . . . . . . . . . . . . . . . . . . . . . 31 2-3 Relationship between beam position and angle of deflection. . . . . . . . . . . . 32 2-4 Free body diagram of beam element dx. . . . . . . . . . . . . . . . . . . . . . . 33 2-5 Vibration amplitude of a free, undamped cantilever tip. . . . . . . . . . . . . . 36 2-6 Vibration amplitude of a free, damped cantilever tip. . . . . . . . . . . . . . . . 37 2-7 Schematic of the extended point probe model. . . . . . . . . . . . . . . . . . . . 39 2-8 Comparison of cantilever magnitude and phase response at resonance between point mass model and flexural beam model . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-9 42 Comparison of cantilever magnitude response between point mass model and flexural beam model slightly off resonance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 2-10 MFM setup for DC current imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2-11 Experimental phase response to varying DC current. . . . . . . . . . . . . . . . . . . 46 2-12 Comparison between phase responses to currents in opposite direction. . . . . . . . . 47 2-13 Comparison between phase responses at different cantilever drive frequencies..... 48 11 LIST OF FIGURES 12 2-14 Experimental phase response to varying DC current. . . . . . . . . . . . . . . . . . . 49 2-15 Experimental phase response to varying lift height. . . . . . . . . . . . . . . . . . . . 50 2-16 Comparison between the experimental and simulated phase responses using the extended monopole model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 2-17 Comparison between the experimental and simulated phase responses using extended dipole m odel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 . . . 55 . . . . . . 56 2-18 Relationship between DC current and maximum to minimum phase difference. 2-19 First and second vibration mode phase imaging of sample magnetic tape. 3-1 Simulated cantilever response to varying AC current . . . . 3-2 Simulated cantilever response to -2 mA AC current . . . . . 3-3 MFM setup for AC current imaging. . . . . . . . . . . . . . 3-4 Circuit diagram for AC current .. . . . . . . . . . . . . . . . 3-5 Experimental magnitude and phase responses to varying AC current.. 3-6 Experimental magnitude and phase responses to varying AC current in the opposite direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7 Magnitude responses to electrostatic force. . . . . . . . . . . 3-8 Schematic drawing of the decomposition method. . . . . . . 3-9 Method for extracting the odd (MFM) component. . . . . . 3-10 Method for extracting the even (EFM) component. . . . . . 3-11 Total and extracted EFM and MFM magnitude responses due to 2 mA AC current. 3-12 Total and extracted EFM and MFM magnitude responses due to 2 mA AC current in the reverse direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13 Extracted EFM and MFM magnitude responses to varying AC current. . . . . . . . 3-14 Extracted EFM and MFM magnitude responses to varying reverse AC current. . . . 3-15 Experimental magnitude and phase responses to varying lift heights. . . . . . . . . . 3-16 Extracted EFM and MFM magnitude responses to varying lift heights. . . . . . . . . 3-17 Comparison between the experimental and simulated magnitude responses using the extended monopole model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-18 Decay in magnetic field and gradient of magnetic field to tip-sample distances. . . . 3-19 Maximum H, and dHz/dz as a function of z. . . . . . . . . . . . . . . . . . . . . . . 3-20 Comparison between experimental and simulated MFM responses. . . . . . . . . . . LIST OF FIGURES 13 3-21 Simulated EFM response to 2 mA AC current. ...................... 80 4-1 Magnitude and phase response at w, = 0. . . . . . . . . . . . . . . . . . . . . . . . . 90 4-2 Cantilever response with no piezo drive. . . . . . . . . . . . . . . . . . . . . . . . . . 91 4-3 Schematic drawing of time-varying parallel-plate capacitor. . . . . . . . . . . . . . . 92 4-4 Magnitude response at varying frequencies. . . . . . . . . . . . . . . . . . . . . . . . 93 4-5 Magnitude responses to varying piezo drive amplitudes when W, + WP = WO. . . . . . 94 4-6 DC current and nondegenerate cantilever response. . . . . . . . . . . . . . . . . . . . 95 4-7 DC current cantilever response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4-8 FFT spectra of output voltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4-9 Non-linearity in input piezo voltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 4-10 Non-linearity in output voltage of a free cantilever. . . . . . . . . . . . . . . . . . . . 98 5-1 Force calibration plot for voltage to metric conversion. . . . . . . . . . . . . . . . . . 103 B-1 Magnetic field produced by a straight conductor at point P. . . . . . . . . . . . . . . 110 B-2 Magnetic field about a point wire. 112 . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-3 Magnetic field about a wire with finite width and height . . . . . . . . . . . . . . . . 112 B-4 The vertical components of H, VH, and V 2 H.. . . . . . . . . . . . . . . . . . . . . . 113 14 LIST OF FIGURES List of Tables 1.1 Table of cantilever parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Table of components used for DC current imaging . . . . . . . . . . . . . . . . . . . 44 3.1 Table of components used for AC current imaging . . . . . . . . . . . . . . . . . . . 61 A.1 Experimental settings for the lock-in amplifier. . . . . . . . . . . . . . . . . . . . . . 106 15 21 16 LIST OF TABLES Chapter 1 Introduction Magnetic force microscopy (MFM) can image magnetic fields down to 10 nm spatial resolution by measuring the interaction between a sharp, magnetized probe tip and the magnetic sample. The technique has become a powerful tool in the magnetic recording head industry, which continually strives for higher density. The integrated circuit (IC) industry, similarly, demands smaller circuits for higher density on a chip, which lowers costs and enables greater portability. However, while there exist highly-developed techniques for voltage measurement on small ICs, current measurement remains a challenging task[6]. Such a technique provides valuable information for quality assurance and failure detection, and the absence of one is costly. From Ampere's Law, charge flow in a conductor results in a "cylindrical" magnetic field about the conductor[13]. By imaging this magnetic field, current measurements can be made. This noninvasive method of current measurement is particularly advantageous because it minimizes the risk of damaging the fragile sample. Thus MFM, with its high spatial resolution, proves to be a good candidate. This thesis explores DC and AC current imaging in MFM theoretically and experimentally. First, an overview of MFM, including its setup, a brief history, and a comparison to other magnetic imaging techniques, is presented in this chapter. A simplified model describing the MFM response to magnetic field is then introduced to provide physical understanding to the magnetic tip-sample interactionq. Its limitations are also discussed, thus motivating the employment of a more complete model. Lastly, an overview of the thesis is provided. 17 Laser Metal wire carrying current Figure 1-1: MFM setup for measuring current. 1.1 1.1.1 Magnetic Force Microscopy Overview MFM is a special application of atomic force microscopy (AFM), which was invented by Binnig et al.[4] in 1986. An atomic force microscope images the sample by scanning a sharp probe tip, attached at the end of a cantilever beam, above the sample. Several operational modes are available for different imaging purposes. Specifically, MFM uses the tapping mode[41], where the cantilever is dynamically driven close to or at one of its resonant frequencies. Interactions between the sample and the tip cause the cantilever to deflect. This deflection is then measured by a pair of photodiodes, which receive the reflection of a collimated laser beam directed at the tip. In the case of MFM, the tip is coated with a ferromagnetic thin film and deflects in the presence of magnetic field. Fig. 1-1 shows the MFM setup of a Digital Instrument Dimension 3000 microscope. The first MFM images were demonstrated in 1987 by Martin et al.[21], who measured forces from a microfabricated magnetic recording head. Individual interdomain boundaries have also been analyzed[10]. Over the years, its high spatial resolution has made it a powerful tool in the data 1.1. 19 MAGNETIC FORCE MICROSCOPY storage industry. Some applications include measurements of the head performance on hard disks and the domain behavior of thin-film and magnetoresistive heads. Outside of data storage measurements, other applications include imaging flux lines in low- and high-T, superconductors[25] and biological samples such as magnetotactic bacteria[27]. Current imaging in MFM was first performed by Campbell et al.[6]. Subsequently, Babcock et al.[2] used current strips to determine tip coercivity and magnetic moment, and Kong et al.[18] similarly quantified magnetic moments using current rings. Gaining a clear understanding of the MFM cantilever response to current necessitates a more detailed introduction of three key components: the mechanical movement of the cantilever beam, the interaction between the MFM tip and a magnetic field, and the setup of the photodetection system. 1.1.2 Movement of Cantilever Beam The AFM cantilever is a miniature elastic beam with characteristic vibration modes and frequencies[30],[34]. A complete solution to the cantilever deflection includes the responses from an infinite set of flexural vibration modes. In general, however, a lock-in amplifier can be used in the measurement process to isolate the cantilever response at a particular frequency. Boundary conditions determine the behavior of the cantilever. The fixed end of the beam is clamped to a bimorph piezoelectric plate, which drives the cantilever. The bimorph piezoelectric plate consists of two pieces of piezo-ceramic that are bonded together so that the differential changes in length of the two pieces due to a difference in voltage given can produce relatively large movements in the cantilever. The other end is attached to the probe tip. In the absence of any tip-sample interaction, the tip-end vibrates freely. Fig. 1-2 illustrates the first four vibration modes of the standard thin film MFM tips from Digital Instruments (type Magnetic Etched Silicon Probe). This model does not include damping and lateral motion from the cantilever. The first four resonant frequencies occur at 89.93, 563.89, 1578.82, and 3093.97 kHz, respectively. The typical material parameters for these cantilevers are shown in Table 1.1, which were obtained from measurements made by Lohau et al.[19] However, if interaction forces are present, then the boundary conditions would differ. This changes the cantilever response. The mathematical model including the presence of DC current is presented and solved in Chap. 2. Chap. 3 contains the theoretical model and solution to AC current imaging. CHAPTER 1. 20 n=1 INTRODUCTION n=2 30 20 T 20 T 10 10 0 0 -10 - 0 0 50 -20 -30' 10 20 2 100 150 x [um] 200 25 0 200 250 -20 0 50 100 150 x [um] 200 -30 2E50 n=3 n=4 30 30 - 20 T 10 E -10 20 T 10 10 0 0 E -10 E-10 -20 -30 -20 0 50 100 150 x [um] 200 250 -30 0 50 150 100 x [um] Figure 1-2: The first four vibration modes, n = 1 to 4, of a freely vibrating cantilever. 1.1.3 MFM Tip An understanding of the MFM tip is required to accurately quantify the magnetic force. It is pyramidal in shape and coated with ferromagnetic thin film, as shown in Fig. 1-3. However, its exact magnetic properties are generally unknown. A simplification of the probe's magnetic behavior to magnetic field is described in the point probe approximation[20], and it has proven to yield satisfactory results. In this model, the force acting on the probe is a function of the effective monopole and dipole moments, which are treated as free parameters to be fitted to the experimental data. 1.1.4 Photodetection System Several methods are available for measuring the cantilever deflection. The photodetection system used in the Dimension 3000 microscope is the slope detection method[23], where a collimated laser 1.1. MAGNETIC FORCE MICROSCOPY Cantilever parameter Length of beam, L Width of beam, a 21 Thickness of beam, b Value 227 28.9 3.37 Unit um um um Young's modulus, E 1.69x10" N/M 2 Density, p Bimorph vibrational amplitude, zo 2330 10 kg/m 3 nm Table 1.1: Table of cantilever parameters beam is focused on the tip and is reflected back into two ajacent photodiodes whose currents are fed into a differential amplifier. It is also called the optical beam deflection technique. This system uses the slope at the tip to determine its deflection, assuming a direct relationship between two, while another system, the optical interferometer technique[21],[31], measures the deflection directly. Fig. 1-4 schematically describes both methods. The two have essentially the same sensitivity[28]. However, the optical beam deflection force microscope has the added advantage of being able to measure lateral forces as well as normal forces[24]. 1.1.5 Comparison to Other Magnetic Imaging Techniques Other magnetic imaging techniques are available and have also been used to measure current. However, a tradeoff between spatial resolution and magnetic field sensitivity separates the different methods to specific applications and needs. Among these techniques, scanning superconducting quantum interference device (SQUID) microscopy[39] offers the best magnetic field sensitivity. SQUIDs are magnetic field detectors that use the relationship between the quantum mechanical phase difference between two closely-spaced superconductors and the supercurrent that flows between them to extract information about the magnetic field. The critical current varies sinusoidally with the integral of the magnetic field through the area of the SQUID loop and the period of modulation is the superconducting flux quantum, h/2e = 2.07x 10-15 Tm 2. Therefore, its sensitivity is very high and increases linearly with the area. Two main disadvantages of scanning SQUID microscopy are that its spatial resolution is limited to about 10 pm and that it requires sensor operation at low temperatures[17]. Scanning Hall probe microscopy (SHPM)[7] consists of a scanning probe attached to a Hall bar. CHAPTER 1. INTRODUCTION 22 Si ti Cantilever beam Thin film coating Figure 1-3: Schematic drawing of a MFM tip (not drawn to scale). It measures the Hall resistance of the bar, which is directly proportional to the magnetic flux through the bar itself. Its sensitivity depends on the proximity of the Hall probe to the sample and the size of the bar. The magnetic field resolution is less than that of scanning SQUID microscopy, but the spatial resolution of the probe is about lpm[26]. Magnetic force microscopy (MFM) has the highest spatial resolution among the magnetic imaging systems, as well as the added advantage of requiring essentially no special sample preparation[12]. However it lacks the magnetic sensitivity that the other systems can provide. Fig. 1-5 compares the three imaging techniques in terms of the two criterion. 1.2 Point-Mass Model An introduction of a simplified model describing the cantilever response is helpful for understanding the system. The point-mass model approximates the cantilever deflection at the tip as a damped harmonic oscillator[34]. The cantilever beam is represented by an effective point mass, meff, attached to a spring with stiffness k,. 1.2. - a) Displacement detection b) Slope detection Figure 1-4: Schematic drawings of the position detection and the slope detection systems. meff is chosen to match the fundamental vibration frequency of the cantilever beam[30], k WO =c meff And kc, which is generally about 2 to 5 N/m, is defined by the physical properties of the beam, which will be explained in Sec. 2.1.4. In the absence of any tip-sample interaction, the cantilever deflection, z(t), is described by meff d 2 z(t) 2 dt fb~t + mefb dt + kz(t) = Foejwt (1.2) where b is the damping term and Foei't is the driving force of the cantilever from the piezo motion. By assuming a solution z(t) = Aeidt, the amplitude and the phase responses of the cantilever can be found: A(w) = (2 Fo/meff -w 2 ) 2 + (WO <b(-) = tan-' where Q = (1.3) )2 ww (1.4) wo/b is the quality factor of the cantilever and is usually about 180. Tip-sample interaction creates an additional force on the cantilever, which can be Taylor-expanded to first order, F(z) = F(zo) + aF(z0 ) (z - zo) (1.5) The zero-order term shifts the equilibrium position of the cantilever tip, but the first-order term EElectron Microscopy 10 -I 10 - 10 - 10_ - - 'kIM F 10 uA I-- >! 10 C 1 uA 0-7 0.1 uA 10- 10 nA 10.9 1010 - 10 10-1 0.01 0.1 0 110-o 0-I(DO--o C i 1 10 100 Spatial Resolution (prn) Figure 1-5: Comparison of magnetic imaging techniques. A tradeoff between magnetic field sensitivity and spatial resolution is shown among scanning superconducting quantum interference device (SSQUID) microscopy, scanning Hall probe microscopy (SHPM), and magnetic force microscopy (MFM). changes the effective spring constant, k' = k, - F' (1.6) where F' = Consequently, the resonant frequency of the oscillator is shifted by 2ke assuming that F' (1.7) < kc. Fig. 1-6 describes the resonant frequency shift, Awo = responses as a result of tip-sample interaction. w' - wo, in both the amplitude and phase 25 1.2. POINT-MASS MODEL 2 12. w' W.0 10- 1 no interaction with mag. force W 0o 1.5 - 8- 0.5 AA / I i - ID "0 6- /WO 0 'D -4 W,0 -0.5 4 -1 - 2- 95 100 Frequency [kHz] 105 95 100 Frequency [kHz] 105 Figure 1-6: The presence of a magnetic field causes a shift in the resonant frequency, Awo = affecting both the amplitude and phase responses. 1.2.1 W, - wo, DC Current Imaging In the presence of DC current, the magnetized tip experiences an external force F, which can be quantified by approximating the tip as either a point dipole or a point monopole. F = V(m -poH) F = qpoH and (1.8) (1.9) where m and q are respectively the effective magnetic dipole and monopole moment of the tip, po is the permeability of free space (47r x 10-1 Wb/Am), and H is the magnetic field produced by the current[12]. Eqn. 1.8 may be easier to understand intuitively by considering the analogous case of an electric dipole in the presence of electric field. In Fig. 1-7(a), the electric dipole rests in an uniform electric field, and the positive and negative ends experience an equal torque. The dipole begins to spin but its center of mass does not move. In the case of a gradient electric field (Fig. 1-7(b)), the end that is in the stronger electric field experiences greater torque and the difference causes the center of mass of the dipole to shift. The concept can similarly apply to a magnetic dipole in a gradient magnetic Electric dipole a) Uniform b) Gradient + electric field electric field Figure 1-7: Illustration of force on an electric dipole. (a) The electric dipole rests in an uniform electric field. There is no net movement. (b) The electric dipole in the presence of a gradient electric field. The end in the stronger electric field experiences greater torque, causing movement. field by imagining the magnetic dipole having a positive magnetic charge on one end and a negative magnetic charge on the other. The monopole approximation, on the other hand, is equivalently an elongated dipole model, where the length of the dipole is so long that only one end experiences the magnetic force. Because the current-induced magnetic field is cylindrical, a first-order approximation of F is sufficient. Thus, Eqn. 1.2 becomes me!! d 2 z(t) dzt eb dz(t) + kz(t) - F'z(t) = FoejWt dt 2 + mt+ dt2 f Jdt (1.10) where F' = az 1)z because only the z-component of the magnetic force affects the cantilever's vertical deflection. The phase change can be quantified, by taking the partial derivative of the original phase with respect to w, evaluated at the resonant frequency, and multiplied by the change in the resonant frequency: 8<> A~b = 9io x Awo Q - VF (1.12) If the tip is approximated as a point dipole, the phase change is as follows: A = Q (m. ioH) (1.13) 27 1.3. FLEXURAL BEAM MODEL And the phase change of the monopole model yields A4) = (q - VpoH) Therefore, the phase response varies with the gradient of the magnetic force. (1.14) If the tip is 2 approximated as a point dipole, the phase responds to V H, and as a point monopole, the phase responds to VH. Both models are used in the DC current imaging theory in Chap. 2 and are compared with experimental results to determine the validity of the two models. 1.3 Flexural Beam Model The point-mass model is a single-mode approximation and fails to model the cantilever response beyond the first mode. Its limitation necessitates the employment of the flexural beam model[30 which includes all the vibrational modes of the cantilever. This enables imaging at the higher-order modes and provides the basis for theory of the non-linear tip-sample interaction between the piezo movement and AC current. The more accurate model also provides a standard with which the point-mass model can compare. The flexural beam model is derived in Chap. 2 and provides the foundation for the current imaging theories. 1.4 Thesis Overview This thesis uses both theoretical and experimental results to explain DC and AC current imaging. Chap. 2 begins by describing the flexural modes of the cantilever beam mathematically. Different methods of modeling the magnetic tip are discussed. A quantitative simulation of the cantilever response to DC current follows. Comparisons to experimental data are shown and discussed. AC current imaging is presented in Chap. 3. Theory is developed using the flexural beam model and experimental results are shown. The chapter concludes with a discussion of sensitivity. Chap. 4 includes a method for non-linear tip-sample interactions and an examination of the intrinsic non-linearity in the AFM system. The non-linear theory is validated by the limit to DC and AC current imaging and used to excite higher-order frequency components. A summary of the thesis is presented in Chap. 5. It concludes with suggestions for future work in the area of current measurement in MFM. 28 CHAPTER 1. INTRODUCTION Chapter 2 DC Current Imaging Before approaching current imaging, models used to quantitatively interpret MFM images must be first presented. In particular, the cantilever beam movement and the tip-sample interaction require detailed examination. As mentioned in Chap. 1, the cantilever vibration can be simplified as a mass-spring model, where the resonant frequency of the system is chosen to match the first vibration frequency of the beam. However, higher-order vibration modes of the beam are neglected, and this model becomes inadequate when imaging at the higher-order modes are of interest or when the higher-order modes are excited. Therefore, a complete model of the cantilever dynamics, the flexural beam model[30], is introduced. A brief description of several models describing the magnetic tip-sample interaction is provided. Reasons for employing the extended monopole model are presented. The two models are first applied to DC current imaging. Simulations at the first resonant frequency are compared with experimental results as well as the point mass model. 2.1 Cantilever Dynamics The cantilever beam is driven in the transverse direction by a piezoelectric bimorph. In order to understand how the tip-end of the cantilever behaves in the presence of magnetic force, first the dynamics of the cantilever movement needs to be discussed. This is described by the transverse beam equation, also known as the Euler-Bernoulli equation. 29 30 CHAPTER 2. DC CURRENT IMAGING dx 2111111 Neutral axis M ( M ArAR d(D x z Figure 2-1: The result of bending moment M exerted on a segment dx of an elastic beam. Tension occurs above the neutral axis and compression occurs below. 2.1.1 Transverse Beam Equation The derivation of the beam equation requires first a relationship between beam geometry and bending moment. Subsequently the two mechanical equilibrium conditions-the sum of the forces and the sum of the moments are zero-are used to arrive at the final equation. The bending moment, M, which is also known as torque to people outside the world of mechanical engineering, is defined as the cross product of force and distance. If it is exerted on both sides of a small segment dx of the beam as shown in Fig. 2-1, a neutral axis emerges-a line above which the beam experiences tension and below which the beam experiences compression. Since strain, c, is defined as elongation over length, it can be described by (R - z)d1 - Rd4 Rd4 _ z R at any distance z from the neutral axis, where R is the radius of bending and d (2.1) is the angle of bending. Because the material of the beam is isotropic, the stress, o, and the strain at any given point are 2.1. CANTILEVER DYNAMICS 31 AA y os AA x z Figure 2-2: Moment, which is also torque, can be expressed as the stress times the distance away from the neutral axis, summed over the cross-sectional area. linearly related: z o = EE = -ER (2.2) where E is the Young's modulus and is in units of N/m2. Stress is defined microscopically as the force per unit area (Fig. 2-2). Since moment is the cross product between force and distance, the total moment on the bending beam can be expressed as the product of the distance away from the neutral axis, z, and stress in the x-direction, summed over the cross-sectional area, M = fA zadA = E z2dA (2.3) z2dA is the moment of inertia of the beam cross-section and is represented by the symbol I. Therefore, moment is EI R (2.4) In order to directly relate moment to the position of the beam, R is defined[35]. From Fig. 2-3, several geometric relations can be deduced. First, since R is the bending radius and arc length is dx z(x) ds ds CD(x) 8 z Figure 2-3: Relationship between beam position and angle of deflection. the product of the bending radius and bending angle, ds = RdO (2.5) tan 4 = dz/dx (2.6) Also, If the beam deflection is assumed to be very small compared to the beam length, then 4 is also very small and justifies the approximation tan - 4. By using the above approximation and combining Eqns. 2.5 and 2.6, the relationship is obtained, 1 R d 2z dx 2 (2.7) and results in a new moment equation, M = EI 2 Z dX2 (2.8) The derivation of the transverse beam equation requires lastly the satisfaction of the mechanical equilibrium conditions[8]. Fig. 2-4 shows a beam element of length dx experiencing an external force, f(x). The shear forces (V and V+dV) and bending moments (M and M+dM) on the two faces are dx f(x)dx M M+dM V V+dV Z Figure 2-4: Free body diagram of beam element dx. In equilibrium, no translation nor rotation occurs. Therefore the net force and the net moment are zero also labelled. The condition of equilibrium at the center of the beam element are + fdx =0 E Fz = (V +dV) -V Y M = (M +dM) - M +(V +dV) dx dx 2 +V 2 =0 2 2 (2.9) (2.10) The sum of the forces are zero in equilibrium because there is no displacement. The net moment, or the net torque, is also zero to ensure no rotation. All except the second term in Eqn. 2.10 causes a torque in the clockwise direction. Therefore, only the second term is negative. As dx approaches zero, the second-order differential terms can be neglected. Therefore, the equations are simplified to dV+ f (x) = 0 (2.11) dx dM d dx + V= 0 (2.12) Finally, if the beam is no longer in equilibrium, then the sum of the forces is no longer zero. It instead equals the inertial force plus the frictional force. The inertial force is defined as the product of mass and acceleration, where mass can be expressed as the material density, p, times the crosssection area, A, and acceleration is expressed as d. The frictional force is defined as the product of the damping term, -y, the mass, pA, and velocity, L. By substituting in Eqn. 2.8 into Eqn. 2.11, the transverse beam equation is obtained, dez dz d2 z EIdx4 +ypAz + pA dt2 = f(x,t) (2.13) CHAPTER 2. DC CURRENT IMAGING 34 In summary, the restoring force from the elastic beam, EI4, works to oppose the external force, f(x, t). When the beam is not in equilibrium, the beam moves. Its velocity is described by the frictional force -ypAd, and its acceleration is given by the inertial force pAp. The sum of the restoring force, the frictional force, and the inertial force, equals the external force. 2.1.2 Free Cantilever without Damping The cantilever motion is first examined without the damping term. In the case of a freely-vibrating cantilever, there is no uniform external force applied on the beam. Therefore, the transverse beam equation is reduced to, EIdx4 + pA dt2 = 0 (2.14) By separating the variables, the solution z(x, t) = Z(x)q(t) can be assumed[9]. Because the piezo drive at x = 0 is proportional to eiwt, by linearity q(t) = Pei"i. Eqn. 2.14 hence becomes EId Z(X) pAZ(x) d2q(t) dX4 q(t) dt 2 W2(2.15) ( This introduces two separate equations: dt 2 -- + w 2 q(t) dd4 Z(X) Substituting the dispersion relation k Z(x) = Alekx = W2 = 0 Z(x) = 0 (2.16) (2.17) and solving Eqn. 2.17 gives a solution for Z(x), + A 2 e--kx + A 3 eikx + A 4 e jkx (2.18) An alternative way of expressing Eqn. 2.18 is Z(x) = C1(cos kx + cosh kx) + C 2 (cos kx - cosh kx) + C 3 (sin kx + sinh kx) + C4 (sin kx - sinh kx) by redefining the coefficients. (2.19) 2.1. CANTILEVER DYNAMICS 35 In order to determine the coefficients C 1, C 2 , C 3 , and C4 , boundary conditions need to be applied. Considering that the cantilever of length L is forced to vibrate from the piezo at the clamped end at x = 0 with the amplitude z(x = 0, t) = a cos wt, the boundary conditions are thus: Z(x = 0) = a dZ(x = 0) = dx d z0 d2 Z(x = L) dx 2 d2 = 00 3 d Z(x = L) =0 dx3 (2.20) (2.21) (2.22) (2.23) is 0 because the beam is clamped to the moving piezo. As derived earlier, d2Zx=L) and rltdt are related to the moment and the shear force at x = L respectively. Because the beam dZ(x=O) 3 d Z( x =L) dx 3 ends at x = L, there cannot be curvature at that point. Also, no shear force is pulling down on the tip-end, allowing the cantilever to vibrate freely. The solution after satisfying the boundary conditions is Z(x)= ((coskx+coshkx)- (sin k L sinh kLL cossincshkL+ kU cosh kU U+ I1 (cos kx - cosh kx) + (sin k Lcosh kL + sinh kL coskLL\ sic cos s cosh + kL (sin kx - sinh kx)) kUL cosh kU + U1I c (2.24) Z(x) peaks as cos kL cosh kL + 1 approaches zero. Therefore, the characteristic equation cos knL coshknL +1 = 0 (2.25) gives the infinite set of wave numbers k, that define the flexural vibration modes, where n is the mode number. Fig. 1-2 in Chap. 1 shows the first four vibration modes of a freely vibrating MFM cantilever. And as illustrated in Fig. 2-5, the amplitude of vibration peaks at the resonant frequencies. CHAPTER 2. DC CURRENT IMAGING 36 160 140 120 100 E 80 80 40 20 0 400 200 600 1000 800 Frequency [kHz] 1200 1400 1600 1800 Figure 2-5: Vibration amplitude of a free, undamped cantilever tip. 2.1.3 Free Cantilever with Damping Inclusion of the damping term, by definition, diminishes the response of the cantilever. The beam equation becomes d4 z dz EIdx4 +YPA d 2z +PAdt2 (2.26) 0 While the spatial distribution of the beam does not change (Z(x) remains the same), Eqn. 2.16 changes, d2q(t) dt 2 dq(t) 2 dt +wq(t)=0 (2.27) Thus the driving frequency is no longer the same as w, and the dispersion relation changes[9]. By defining the driving frequency as Wdi and the quality factor as Wdi Q = , a relationship between and the original w can be obtained: Wdi 2 where w = k +I (2.28) EI Fig. 2-6 shows the damping effect in terms of vibration amplitude as a function of driving frequency. The value of Q needs to be determined experimentally by the equation Q = fo/Afo where fo is the resonant frequency and Afo is the half-width frequency. According to the cantilever tuning fork routine on the DI 3000, Q of the MESP-HM tips at the first resonant frequency was 2.1. CANTILEVER DYNAMICS 37 6- 5- 4- 2- 200 400 600 800 1000 Frequency [kHz] 1200 1400 1600 1800 Figure 2-6: Vibration amplitude of a free, damped cantilever tip. Q at the second resonant frequency is more varying and is found to be between 450 and 550. The model above assumes that Q is 180 at the first resonant frequency and 500 at the found to be 180± 10. second. 2.1.4 Comparison to the Point-Mass Model As mentioned in Sec. 1.2, the cantilever dynamics can be approximated by a mass-spring model. The spring constant, kc, is defined as F k -I (2.29) and can be expressed in terms of the beam parameters[34]. If there is a downward force F pushing on a point a on the beam, the moment at any point x is M = F - (x - a) (2.30) or equivalently, force times distance. By combining Eqn. 2.8 and Eqn. 2.30, a relationship between F and z can be found, z = 6EI x 2 (x - 3a) (2.31) The point of interest is at x = L, assuming that a = L as well. Therefore, an expression for k, CHAPTER 2. DC CURRENT IMAGING 38 is derived: ke = 3E1 (2.32) The resonant frequency of the point-mass model is determined by the fundamental vibration mode of the cantilever beam, and the fundamental mode occurs at kiL = 1.875 according to Eqn. 2.25. Therefore by using the dispersion relation, WO 2.2 = (1.875)2 E (2.33) Magnetic Tip-Sample Interaction The force on a tip due to the magnetic field of the sample can be represented as a convolution between the probe moment and the sample stray field for a particular tip-sample separation. However, while the sample stray field is generally known, the tip magnetization, which originates from a ferromagnetic thin film coating on a pyramidal silicon tip (see Fig. 1-3), may be irregular in shape and is much less predictable. Furthermore, tip hysteresis loops and coercivities are usually unknown and there may be strong interactions between the magnetization of the sample and the tip. Current models describing the magnetic force on a tip have so far restricted themselves to the hard magnetic case, where the magnetization of the sample and the tip remains undisturbed during the scanning process, although the field-dependence of MFM probes have been investigated experimentally[2]. The point dipole approximation, first proposed by Mamin et al.[20] describes the tip as an ideal dipole, with magnetic dipole moment m. A slightly more complicated model describes the tip as the sum of an ideal monopole and an ideal dipole, where the monopole is an approximation of an elongated dipole. The force experienced by the tip due to magnetic field is, F = po(q + m - V)H (2.34) where q is the magnetic monopole moment of the tip and H is the magnetic stray field from the sample[14]. Several groups have experimentally determined the effective values of q and m for various MFM tips[2],[32]. Kong et al.[18] used microfabricated current rings, to take advantage of well-defined MFM tip magnetic pole d- - sample Figure 2-7: Schematic of the extended point probe model. magnetic fields produced by the rings. They presented values for the monopole (2.8x 10-6 emu/cm) and dipole (3.8 x 10-9 emu) moment of a tip covered with 65 nm cobalt for a current ring with 5pm diameter and found that the dipole moment value when using a 1pLm diameter ring is about one order of magnitude smaller. Lohau et al.[19] explained the difference between the two rings by relating a characteristic decay length of the magnetic field to an effective volume of the magnetic tip that is actually interacting with the field. They concluded that the usage of the point probe approximation needs to be adjusted by an effective distance, 6, above the tip apex, because the location of the point probe within the real physical tip depends on the magnetic stray field. Similar observations have been made by Mamin et al.[20] for the dipole model and Belliard et al.[3] for the monopole model. Fig. 2-7 shows the schematic of the extended point probe model. The lift height, d, is a fixed distance above the sample where the tip scans and is manually controlled by the user, and the point probe is effectively at a height z = d + 6 above the sample. Lohau et al. also determined that "an unambiguous MFM-image analysis can only be performed when using either the monopole or the dipole contribution of the magnetic tip." Similar conclusions were made by van Schendel et al.[16],[38], who used transfer functions in the Fourier domain to quantify the magnetic force on the tip. Calculations are simplified in Fourier space because transfer functions are multiplied instead of convoluted. When compared to the point-pole tip models, the authors found the extended monopole model to agree best with their simulations. While the dipole and monopole models only match the ex- CHAPTER 2. DC CURRENT IMAGING 40 perimentally determined tip surface charge distribution at specific spatial wavelengths of sample magnetization, the extended monopole model matches over a greater range of wavelengths. 2.3 Theory In the presence of a device carrying DC current, the tip-end of the cantilever experiences an attraction or a repulsion force in the z-direction. This force can be expanded into a Taylor series, F(z) = F(zo) + OF 1 a2 F 1. (z - zo) + 2z 2 Iz (z - zo) 2 +... The first term shifts the equilibrium position of the cantilever beam at x (2.35) L. The excition force around the equilibrium can be approximated by the second term, which can be represented as an effective spring with stiffness k*[29], where: k* = 9F(z)| (2.36) Oz Therefore, the boundary condition at x = L changes while the cantilever, clamped to the piezo, is still driven at x = 0. The moment remains zero at x = L because nothing is attached beyond that end to create a curvature. However, the shear force now equals the magnetic force[30], El dx 3 = k*Z(x = (2.37) L) Z(x) can again be found by using the four boundary conditions to solve for the coefficients of the general solution (Eqn. 2.19): Z(x) = 2 ((coskx+coshkx)M sin kL sinh k L-cosh kL sin kL-cos kL sinh kL cos kL sinh k L-cosh kL sin kL-M(cos kL cosh kL+1) ' 2 cos kL cosh kL-M(cos kL sinh kL+cosh kL sin kL) cos kL sinh k L-cosh kL sin kL-M(cos kL cosh kL+1) where M = (cos kx (sin kx sk - cosh - kkx) + sinh kx(23 3 Ek. The characteristic equation that defines the resonant frequencies is calculated in the same way 2.3. THEORY 41 as for a free cantilever. Z(x) approaches infinity as the terms in the denominator approach zero: cos k, L sinh k, L - cosh k, L sin k, L = E Ik3 k*,3 (cos k, L cosh k, L + 1) (2.39) As k* approaches zero, in the limit that the spring is infinitely soft, the characteristic equation reduces down to the same for a free cantilever. And if the spring is infinitely hard, k* = 00 and the characteristic equation becomes the same as in the case of a pinned end[30]. The concept of the force on the tip shifting the resonant frequency of the cantilever, as described in the point mass model, is shown here mathematically. Althernatively, E can be expressed as kc(k1) . Depending on the effective stiffness of the external force k* compared to the effective spring 3 constant of the beam k,, the resonant frequency ranges between the limits of a free cantilever and a pinned cantilever. At the first vibrational mode, kiL of a free cantilever is 1.8751, kiL of a pinned cantilever is 3.9266. If L! = 0.1, kiL = 1.91891, and if j- = 10, kiL = 3.1677[30]. By using the extended monopole model, the magnetic force is F = qpoH (2.40) where H of DC current flowing through an infinitely long wire with finite height and width is derived in App. B and the effective distance between the sample and the monopole is z = d + 6. Because the tip is specifically magnetized in the z direction, the z component of H dominates the magnetic tip-sample interaction. Therefore, Fz = qpoHz (2.41) This means that the magnetic force is proportional to the magnetic field. From the point mass model, it was shown that the magnitude response is proportional to the magnetic force and that the phase response is proportional to the gradient of the force in the z-direction, providing a better intuitive understanding of the effects of the magnetic force on the tip behavior. Therefore, a comparison of the cantilever's responses between the two models is necessary. 2.3.1 Comparison to the Point-Mass Model at the First Vibration Mode The cantilever response at the first vibration mode, using both the point mass model (PMM) and the flexural beam model (FBM), is shown in Fig. 2-8. The cantilever, using the parameters given in CHAPTER 2. DC CURRENT IMAGING 42 Table 1.1, is driven at its fundamental frequency (89.9 kHz) where the maximum phase response is expected. The wire used in the simulation is 2pm wide and 200 nm tall, and q and z are assumed to be 10-6 Am and 400 nm respectively. The two models yielded very similar results. Furthermore, the phase of the deflection response, Z(L), and the slope response dZ(L)/dx of the flexural beam model were found to be identical. 2820 2800-C2780 2760 ' 2740 2272027002680 -10 -8 -6 -4 -2 I 0 x [um] 2 I 20 4 6 I 8 10 I 10- 0 -10 -10 -8 -6 -4 -2 0 x [um] 2 4 6 8 10 Figure 2-8: Comparison of cantilever magnitude and phase response, at resonance, between point mass model (PMM) and flexural beam model (FBM) along a 2pm-wide wire centered at x = 0, carrying 20 mA DC current. For the most sensitive magnitude response, the cantilever is not driven at the first resonant frequency[34]. Instead, it is driven at the frequency that has the highest magnitude slope, which is slightly off the resonant frequency. The magnitude response at this frequency is shown in Fig. 2-9. In these simulations, this frequency is determined to be at 89.71 kHz. It is evident that the two models also behave very similarly at the fundamental mode. The magnitude response of dZ(L)/dx is plotted separately because it yields a larger response. 2.4. EXPERIMENT 43 2.6 2.5 2.3 'E2.2 22.1 2 10 1.55 -8 -6 -4 -2 -8 -6 -4 -2 0 x [um] 2 4 6 8 10 0 2 4 6 8 10 X 104 1.5 1.45 1.4 1.35 2 1.3 1.25 1.2' -1c 0 x[um] Figure 2-9: (Top) Comparison of cantilever magnitude response, slightly off resonance, between point mass model (PMM) and flexural beam model (FBM) along a 2pm-wide wire centered at x = 0, carrying 20 mA DC current. (Bottom) Magnitude slope response of the cantilever. 2.4 2.4.1 Experiment Setup The experimental setup for DC current imaging is shown in Fig. 2-10 and the components are listed in Table 2.1. The experiments were performed at room temperature. At the beginning of each session, the tips are calibrated by first imaging a magnetic tape sample, which was from Digital Instruments. To optimize scanning sensitivity, the tip is scanned very slowly across the sample, at a rate of 0.05 Hz. Data acquisition was performed external to the AFM, using a lock-in amplifier along with LabVIEW and Matlab code. The data sampling rate of the lock-in was set at 32 Hz and the time constant was at 30 ms. Two output signal from the AFM's signal access module are available: RMS amplitude (IN 0) and vertical deflection (AUX A). The vertical deflection output was unpredictable during DC current imaging sessions, sometimes showing strange slants in the response. Thus, the Laser E f '\ Lock-in Amplifier S A Figure 2-10: MFM setup for DC current imaging. The function generator drives the cantilever at frequency f and provides the reference signal for the lock-in amplifier. The input signal to the lock-in is taken from the photodetector signal of the AFM. Component Function Generator DC Power Supply Lock-in Amplifier Atomic Force Microscope MFM tip Vendor Agilent Agilent Stanford Research Systems Digital Instruments Digital Instruments Part Number 33120A E3633A SR844 Dimension 3000 MESP-HM Table 2.1: Table of components used for DC current imaging signal was taken from the RMS amplitude output, which is also the feedback signal. Because the feedback system is amplitude-modulated, the magnitude of the feedback signal is not believed to represent the direct magnitude of tip deflection. The phase response, on the other hand, should accurately represent the phase of the cantilever vibration. The wire sample was fabricated by Mathew Abraham. The fabrication process as well as the data acquistion steps are explained in App. A. 2.4. EXPERIMENT 2.4.2 45 Results By driving the cantilever at the first resonant frequency of the tip, the phase response to varying current was obtained. Fig. 2-11 shows the raw data acquired by the lock-in, the normalized data, and a plot of the difference between the maximum and minimum phase response as a function of current. The normalized data is centered at its mean and provides a better comparison among the responses at different current levels. It is the change in the phase response that indicates the change in the effective spring constant by DC magneitc field. The increased phase response due to greater current is evident. A line-fit through the third plots yields a slope of 0.118 deg/mA. The resonant frequency of the tip was 59.44 kHz. The wire used in the experiment is 2pim wide and 200 nm tall. The lift height between the sample and the tip was 200 nm. This means that the tip is first scanned right above the sample, obtaining good topographical data. As a second pass, the tip is lifted by an amount specified by the user and by incorporating the topographical data, the tip-sample distance is maintained throughout second scan. One disadvantage to driving the cantilever externally is a tradeoff between topographically accuracy and magnetic field sensitivity. The setup of the AFM's internal piezo driver allows the topographical and lift scans to be performed at different tip drive frequencies as well as tip drive amplitudes. Therefore, the topographical scan can occur at the frequency of optimum magnitude response, which is usually 200-300 Hz off the resonant frequency, and the lift scan can occur on the resonance, giving maximum phase response. However, by applying the cantilever drive externally, we are limited to one frequency. The topographic effects on phase response are shown in Fig. 2-12. Sync outputs generated by the DI instrument were available inside the controller machine. However, the signal had noticeable delay which prevented accurate synchronization to the start of each scan. As the tip scans across the sample in the lift mode, it is raised by 200 nm (the height of the wire) upon reaching the wire and again lowered by the same amount once it passes the wire. Because the maximum phase response also occurs near the wire edges, the signals can be contaminated if topography wasn't well-tracked and the tip does not lift and lower right at the edge of the wire. In the case where the magnetic field is in the counter-clockwise direction, the tip lowered before it hit the edge of the wire and thus the phase response was amplified. Similarly, when the magnetic field is clockwise about the wire, the tip lifted before reaching the wire and therefore diminished the response to magnetic field. The tip appears to be more susceptible to topographical contamination in the presence of a CHAPTER 2. DC CURRENT IMAGING 46 b) a) 82.5 r 82 1 0.5 (5; 4)81.5 0 C) Ca IL IL -0.5 81 80.5 -1 80 -10 0 -5 -5 0 10 5 -10 -5 x [um] C) 0 x [um] 5 10 2 -8 1.5 C) U . c15 0.5 (L 0 10 5 15 I [mA] Figure 2-11: Experimental phase response to varying DC current flowing through a 2pim-wide wire centered at x = 0. The cantilever is driven on resonance. Current levels are 0, 0.5, 1, 2, 3, 4, 5, 7, 9, 10, 12, and 15 mA. The arrows are drawn in the direction of increasing current. Tip-sample distance was 200 nm. a) Raw data from the lock-in. b) Data normalized by the mean. c) The difference between maximum and minimum phase response. 2.4. EXPERIMENT 0.3 47 0.6 3 mA 7 mA 0.2 0.4 0.1 '53 0) a) Cz 0 -0.3 0 U) -0.1 -0.2 0.2 (D _r_ -0.2Cz CL CW magnetic field -0.4 -1 0 -5 0 x[um] -0.4- CCW magnetic field 5 -0.6-0.8 -1 0 10 0.6 0.4 5; a) Ca 1 9 mA -5 0 x[um] 5 10 0 x[um] 5 10 12 mA 0.5 0.2 0 0 -0.2 C .C - a_ -0.4 -0.5 -0.6 -0.8' -1 0 -5 0 x[um] ' 5 '-1 10 1 -10 -5 Figure 2-12: Comparison between phase responses to currents in opposite direction. distance is 200 nm Tip-sample CHAPTER 2. DC CURRENT IMAGING 48 repulsive magnetic field. The phase responses of the two current directions in attractive magnetic fields are very similar. To compensate, the tip was driven slightly off its resonance, at 59.41 kHz, to obtain reasonable topographical information as well as good sensitivity. Fig. 2-13 shows the results compared to driving the tip on its resonance. The response is improved in the repulsive magnetic force regime and confirmed in the attractive magnetic force regime. Fig. 2-14 shows the results of driving the cantilever off resonance. A line-fit of the phase difference response to current gives a slope of 1.06 deg/mA. As expected, the phase sensitivity is slightly less than if the cantilever were driven on resonance. However, the results are more reliable. Some topographical artifacts are believed to have affected the phase reponse. As seen in the case where there is no current flowing through the wire, some phase response is still detected and it corresponds to the topography of the wire. This effect can be reduced by sensing at a higher lift height. 0.4 0.6 5 mA 7 mA - 0.4 Detuned -- Resonant 0.2 0.2 Cu a.CO 0 CCu3 (L (L 0 -0.2 -0.2 -0.4 -0.41 0 -5 0 x [um] 5 -10 10 0.6- 0.4- 1 9 mA -5 0 x [um] 5 10 -5 0 x [um] 5 10 12 mA 0.5 0.2Cu -o PA0 Cu .C Cu C-0.2 0 Cu a-0.4 -0.5- -0.6 -0.8 -0 -5 0 x [um] 5 10 -10 Figure 2-13: Comparison between phase responses to cantilever driven on (59.44 kHz) and off resonance (59.41 kHz). Tip-sample distance is 200 nm. 2.4. EXPERIMENT a) 49 b) 92 91.5 1r 0.51 91 0 [L 90.5 -C IL -0.5 90 89.5-10 -5 ' 5 0 x [um] -1' -1 0 '0 10 -5 0 x [um] 5 10 2 C) -~1. 5 C 6 a) *~0.5 00L 10 5 15 I[mA] Figure 2-14: Experimental phases response to varying DC current flowing through a 2pm-wide wire centered at x = 0. The cantilever is driven off resonances, at 59.41 kHz. Current levels are 0, 0.5, 1, 2, 3, 4, 5, 7, 9, 10, 12, and 15 mA. The arrows are drawn in the direction of increasing current. Tip-sample distance was 200 nm. a) Raw data from the lock-in. b) Data normalized by the mean. c) The difference between maximum and minimum phase response. CHAPTER 2. DC CURRENT IMAGING 50 a) 8 2 .5 r b) 82 1 0.5 81 .5 .D 0 IL Ca 81 -0.5 80.51 80'-10 -5 5 0 x [um] C) -1 -1 10 0 -5 0 x [um] 5 10 2 1.8 a, 2.1.6 1.4 1.2 6\ 0.8 200 400 O 600 800 Lift Height [nm] 0 1000 Figure 2-15: Experimental phases response to varying lift height. Current level is 15 mA. The cantilever is driven on resonances. Lift heights are 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, and 900 nm. The arrows are drawn in the direction of increasing lift height. a) Raw data from the lock-in. b) Data normalized by the mean. c) The difference between maximum and minimum phase response. 2.4. EXPERIMENT 2.4.3 51 Comparison to Theory at the First Vibration Mode The extended monopole (Fig.2-16) and dipole (Fig. 2-17) models are both compared to the detuned experimental results with 15 mA DC current. For the extended monopole model, q and 6 are fitted to match the experimental results. The values obtained for q and J are 1.9x 107 Am and 650 nm, given that the lift height is 200 nm. The experimental and simulated results are closely-matched, particularly within the the width of the wire, which is between x = -1p1m and x = 1pm. Further away from the wire, the experimental results fall off slower than the simulated. This is possible due to the physical volume of the tip, which is not simulated in the monopole model. As the tip moves away from the wire, the outer rim of the pyramidal tip may still interact with the magnetic field, causing some phase response. The extended dipole model simulation agrees less-well with the experimental result. With 15 mA current, m, and 6 are found to be 2.5x10- Am 2 and 1000 nm, respectively. However, away from the edge of the wire, the phase response of the extended dipole model falls off faster and even overshoots a little before approaching zero. Therefore, it appears that the extended monopole model better represents the magnetic tip-sampe interaction. In order to approximate a reasonable 6 for the extended monopole model, the maximum to minimum phase response is plotted along a linear increase in current (see Fig. 2-18). Different lift height values are examined. They range from 200 nm to 1000 nm, incrementing by 100 nm. According to (c) in Fig. 2-14, the phase difference varies linearly with current. In Fig. 2-18, a linear dependence begins to emerge around lift height z = 700 nm, justifying the need to include an effective distance, 6, to the physical tip-sample distance. 2.4.4 Second Vibration Mode with Magnetic Tape In addition to modeling the cantilever response at the fundamental frequency, the flexural beam model offers the flexibility of examining the responses at higher-order cantilever vibration modes. The vibration modes of a free cantilever are determined by Eq. 2.25 and the first five kaLs are 1.875, 4.694, 7.855, 10.996, and 14.137, respectively. To image at the second vibration mode, the second resonant frequency is calculated. Using the same cantilever parameters given in Table 1.1, the calculated value for k2L, and Eq. 2.28, the second resonant frequency, f2, is found to be 563 kHz. It is about 6.26 times larger than the first resonant frequency, fi. CHAPTER 2. DC CURRENT IMAGING 52 With this information, as well as the experimentally determined values for Q at the first and the second resonant frequencies, the point mass model can be used to find the phase response at the second vibration mode the same way it calculated the response at the first. Assuming the effective mass, meff, of the tip remains constant, a factor of 6.26 increase in the frequency corresponds to a factor of 6.262 increase in the effective spring constant. Q, from experimental data, increases by a factor of 2.37 to 3.24. Therefore, according to Eq. 1.13, the phase response at the second vibration mode should be 12 to 16 times smaller than the phase response at the first vibration mode. This decrease in sensitivity prevents accurate DC current imaging at the second mode. However, we did image magnetic tape with the second mode, because the tape has a stronger magnetic field. Fig. 2-19 shows the results. The data was taken straight from the DI instrument. 6 The numeric values obtained from the ASCII export function require data conversion: x/21 x 180 where x represents the phase values. The measured first and second resonant frequencies of the tip are 62.86 and 399.96 kHz, respectively. The increase in Q at the first is 184.89, and it is 533.28 at the second. the Q and the frequency from the first to the second There is a factor of 2.88 and 6.36 vibration mode. Therefore, from the theory, a factor of 14.03 decrease in the phase response is expected. The experimental phase response at the second mode was found to be 15.68 times less than the first. The flexural beam model is needed to model imaging at the high-order vibration modes. However, when the magnetic field from the sample is non-time-varying, the point mass model suffices once the high-order resonant frequency is calculated. When driving at the first and second modes, the drive amplitudes should be tuned accordingly to ensure the free cantilever vibrational amplitudes are equivalent. It serves as an experimental control for a more precise comparison between the responses at the two modes. The drive amplitudes vary with the tip as well as the placement of the tip on the cantilever holder and can only be determined experimentally. 2.5 Summary This chapter covers the derivation of the flexural beam model, the different proposals for the modeling magnetic force on the tip, the theoretical and experimental results of DC current imaging at the first vibrational mode, and imaging of the magnetic tape at the second mode. 2.5. SUMMARY 53 Three key points are summarized. First, by matching the experimental results to both the extended monopole and dipole model, we found the extended monopole model to better represent the magnetic tip-sample interaction. The effective monopole moment and displacement for 15 mA DC current were found to be 1.9x 10-7 Am and 850 nm respectively. According to the point mass model, the phase of the cantilever vibration responds to the gradient of the magnetic field created by the DC current. Secondly, the overall magnetic imaging sensitivity at the second harmonic is worst than the first, even thought the Q value at the second harmonic is higher. The phase response is proportional to the ratio of Q over the effective spring constant. While spring constant increases even more, by a factor of Q increases (W2/w1) 2 . at the second mode, the effective Therefore, the overall phase response at the second mode is less. Third, sensitivity in DC current imaging was determined to be at 1 mA. One possible limiting factor is thermal noise. In Chap. 3, AC current imaging is discussed. The time-varying current acts as a driving source and the piezo no longer needs to drive the cantilever. CHAPTER 2. DC CURRENT IMAGING 54 0.80.60.40.2 0 ----------------------0.2 -0.4q = 1.9xo10- Am LH =850 nm -0.6-0.8-1111 -10 -8 -4 -6 -2 0 x [um] 2 4 8 6 10 Figure 2-16: Comparison between the experimental and simulated results of the cantilever phase response to a 2 pm-wide wire carrying 15 mA DC current. The simulation used the extended 850 monopole model to describe the magnetic tip-sample interaction. q = 1.9xiO- 7 Am and z nm give the best fit result. per -- - 1 nal 0.80.60.4 - 0.2 0 ---- - - -/--- - - - ~ -0.2 0.4 - -- m =2.5x104 Am2 LH = 1200 nm -0.6-0.8- -10 -8 -6 -4 -2 0 x [UM] 2 4 , 8 10 Figure 2-17: Comparison between the experimental and simulated results of the cantilever phase response to a 2 pm-wide carrying 15 mA DC current. The simulation used the extended dipole model to describe the magnetic tip-sample interaction. mz = 2.5 x 10-13 Am 2 and z = 1200 nm give the best fit result. 2.5. SUMMARY 55 3- 2.5- 2 S1.5- _r- Ca 0.5 - 0 5 10 15 I [mA] Figure 2-18: Relationship between DC current and maximum to minimum phase difference at varying liftheights. The extended monopole model is used. Lift heights range from 200 nm to 1000 nm, incrementing by 100 nm. The arrow is in the direction of increasing lift heights. CHAPTER 2. DC CURRENT IMAGING 56 a) b) 15 0.8 0.6 10 0.4 5 o 0.2 0 a a. U) 0 0 -0.2 -5 -0.4 II -0.6 ) 0 3 2 c) 2 1 0 3 x [um] x [um] 15 First mode 10 -(Second mode) x 15.68 D5 "a .c 0 -5 -10 0 2 1 3 x [um] Figure 2-19: First (a) and second (b) vibration mode phase imaging of sample magnetic tape (3 ym). (c) Best fit of second mode imaging to first mode imaging. Chapter 3 AC Current Imaging The theory and experimental results for AC current imaging are presented in this chapter. Although the tip-sample force is now time-varying, the system remains linear by reducing the piezo drive to zero and limiting the driving force to the magnetic field created by AC current. Magnetic force from a time-varying current is first derived and applied to the boundary conditions. Simulations of the cantilever response are presented. Secondly, experimental results are shown and compared to the theory. 3.1 Theory The flow of AC current instead of DC current through the wire sample introduces a new frequency component in the beam vibration. The AC magnetic field produced by the current acts as a new driving force on the cantilever. As long as the current frequency is different from the piezo drive frequency, the linear cantilever response at the current frequency can be filtered out by a lock-in amplifier. 3.1.1 Magnetic Force The derivation of the magnetic force is similar to the DC case. Referring to App. B, the current flowing through the conductor is now a product of an amplitude, I, and a time-varying term, e3'. The new term is linearly carried into the expression of the magnetic field. The z-component of the 57 CHAPTER 3. AC CURRENT IMAGING 58 magnetic field with AC current flowing through a wire with finite width and height is thus, Ieswct HZ = 27rW I H f 0 W/2 W2 w/2 0zO - H (X- _ (3.1) )dzdx XO p2 + (Z -~zo)2 Since the approximate magnetic force from the extended monopole model is linearly related to the magnetic field, it also can be expressed as the product of the force from the DC current and eJwt. 3.1.2 Boundary Conditions When using Taylor series expansion to approximate the magnetic force, the first term, F(zo), represents the new driving force. Therefore, it can no longer be neglected. Furthermore, it is orders of magnitude larger than the higher-order terms. Therefore, only the first term in the Taylor series is included. At x = 0, the cantilever beam is clamped. Its position and first derivative are both zero. Still no moment is applied at x = L. However, the shear force there equals the magnetic force. The new boundary conditions are as follows: z(X = 0,t) = 0 (3.2) &z(x=,t) = 0 (3.3) ElEI 2z(x,t)= OX2Lt- (.4 (3.4) El az(x=L,t) = Fz(zo)eJowt (3.5) where w, is the current frequency, F(zo) is the same Taylor series component as in the DC case since the time-varying term is added externally, and zo is the effective tip-sample distance. 3.1.3 Solution The solution is assumed to take the form of z(x, t) = Z(x)ewct, where a general solution for Z(x) is given in Eqn. 2.19. After applying the boundary conditions, Z(x) is solved: Z(x) = 1 ( sin kL+sinhkL M os kLcosh kL+1 (COS k (co ,( - cosh kx) - k)~~ cos kL+coshL coskLcoshkL+i (sin k snx - sinh k) ihkx (3.6) 36 3.1. THEORY 59 and wc = k2 where M = ( 1 - + 2). Fig. 3-1 shows the simulated cantilever response to varying AC current flowing through a 2 umwide wire. The current values are 0.5, 1, 1.5, and 2 mA, with the arrow pointing in the direction of increasing current. The current frequency matches the first vibration mode of the beam. The effective monopole moment and tip-sample distance are: q = 8x10-6 Am and z = 1 jm. The discontinuities at x = -1 jm and x = 1 jm are a result of the constant lift height between the tip and the top of the sample. 70 6050 -40 - '30 - 220 100 -30 -20 -10 -20 -10 0 x [um] 10 20 30 100 50- 0 Co -50- -100' -30 I 0I.___________ 10 x [um] __L_______ 20 30 Figure 3-1: Simulated cantilever magnitude and phase response to AC current at the first beam vibration mode. The current values range from 0.5 to 2 mA, in increments of 0.5 mA. The arrow points in the direction of increasing current. The wire is 2 pm-wide and centered at x = 0. q 8x10- 6 Am, z = 1 pm. The magnitude response increases linearly with current. It is symmetric about the center of the wire and gradually falls off to zero as the tip travels further away from the wire. The phase response, on the other hand, does not vary with current. But it shifts 180 degrees as the tip crosses the middle of the wire. On one side of the wire, if the tip experiences a "push-pull" force from the AC magnetic field, then it should experience instead a "pull-push" force when it is on the other side, because the z-component of the magnetic field switches sign at the middle of the wire. CHAPTER 3. AC CURRENT IMAGING 60 When the current is in the reverse direction, the magnitude response remains the same since it is symmetric, but the phase is flipped (see Fig. 3-2). Because the current is now going in the opposite direction, the "push-pull" effect on the tip reverses. 70 60-50 40 "E30-- 220-10 --30 -20 -10 0 10 20 30 10 20 30 x [um] 100 50- 0. -50- -1n II -30 -20 -10 0 x[um] Figure 3-2: Simulated cantilever magnitude and phase response to -2 mA AC current at the first beam vibration mode. The wire is 2 pm-wide and centered at x = 0. q = 8x10- 6 Am, z = 1 pm. Contrary to DC current imaging, the cantilever response to AC current is mainly contained in the magnitude data. Therefore, the cantilever vibrates in direct response to the magnetic force, and not its gradient. When using the extended monopole model, we can conclude that the cantilever responds directly to AC magnetic field while it responds to the gradient (in z) of DC magnetic field. 3.2 Experiment Experimental results of AC current imaging are shown in this section. The experimental setup is first explained. The magnitude and phase responses to varying current levels and directions are presented. The asymmetric magnitude responses lead to the hypothesis of electrostatic force interference. A method for decomposing the EFM and MFM response from the magnitude data is developed and the hypothesis is verified. Laser ; Lock-!i.....n Amp IIffier: A Function generator Figure 3-3: MFM setup for AC current imaging. The current is driven by an external function generator at frequency we, which becomes the only driving force on the cantilever. The lock-in amplifier receives signal from the AFM photodetector. 3.2.1 Setup The components used in the AC current imaging experiments are listed in Table 3.1 and the setup is shown in Fig. 3-3. Component Vendor Part Number Function Generator Agilent 33120A Lock-in Amplifier Stanford Research Systems SR844 Atomic Force Microscope Digital Instruments Dimension 3000 MFM tip Digital Instruments MESP-HM Table 3.1: Table of components used for AC current imaging As mentioned earlier, the DI instrument uses the LiftMode technology to sense long-range forces, such as electrostatic and magnetic forces. LiftMode allows the tip to first perform a topographical scan across the sample, in which the tip is positioned close to the sample surface and amplitude feedback is used to extract the topographical makeup of the sample. Once topography is known and Function Generator 50Q I I + I1 II II II |i I I I I_ _ _ _ _ _ _ Wire Sample Figure 3-4: Circuit diagram for AC current. The output impedance of the function generator is 50 Q. Two wires on the sample are bonded to the contact pads, each has about 100 Q of resistance. stored into memory, the tip rescans across the sample at some fixed distance above the topography to ensure that the tip is responding to the long-range forces and not contaminated by the topography. In the AC current experiments, the tip is first driven at a frequency where the magnitude response is most sensitive, in order to acquire good topographical data. During the lift scan, the piezo drive amplitude and frequency are both set to zero, leaving the AC magnetic field as the only driving force. The current is driven externally by the function generator at frequency w,. The sync output from the function generator provides the reference signal for the lock-in amplifier. The AFM photodetector signal is taken from the vertical deflection output (AUX A) of the Signal Access Module (SAM) and fed to the lock-in as the input signal. Fig. 3-4 shows the circuit diagram between the voltage from the function generator and the wire sample. The output impedance of the function generator is 50 Q. Only two wires are bonded to the contact pads and each has about 100 Q of resistance. Therefore, the amplitude of current passing through one of the wires is vi, x 100. With the 2 pm-wide wire in the middle, the tip scans a distance of 50 Pm, across the wire, at a rate of 0.1 Hz (5 pm/s). By the use of LabVIEW, the lock-in samples the input signal at 32 Hz. The data is stored into an internal buffer and subsequently downloaded to a computer and read via Matlab code. The details of the data acquisition process is given in App. A. In order to minimize the effects of the 12-degree incline between the cantilever and the sample 63 3.3. RESULTS surface, the tip should scan symmetrically across the wire. This is accomplished by positioning the long-dimension of the wire sample parallel to the long-dimension of the cantilever and setting the scan angle to 90 degrees. 3.3 Results Figs. 3-5 and 3-6 show the experimental results to varying AC current in both directions. The lift height was fixed at 200 nm and the wire is 2 1am-wide, centered at x = 0. A 180-degree shift in the phase response is observed, and it agrees with theory. The offset in the actual experimental phase values could be due to the AFM's internal circuitry or delay between the photodetector output signal to the lock-in. 0.06 0.05 '0.04 0.030.02 0.01 -25 -20 -15 -10 -5 -20 -15 -10 -5 0 5 10 15 20 25 0 5 10 15 20 25 200 100- 0- - 100 - -25 x lum] Figure 3-5: Experimental magnitude and phase responses to varying AC current through a 2pmwide wire centered at x = 0. The cantilever is driven on resonance. Current levels are 5, 25, 50, 100, 250, 500 pA and 1, 2, 3, 4 mA. The arrows are drawn in the direction of increasing current. Tip-sample distance was 200 nm. The magnitude response, however, does not follow the simple AC current imaging theory. While the theory predicted a symmetric magnitude response about the center of the wire, the experimental results were asymmetric. CHAPTER 3. AC CURRENT IMAGING 64 OX 8 0.061 57 0.041 CO 0.02 0 -25 -20 -15 -10 -5 0 5 10 15 20 25 -20 -15 -10 -5 0 x [um] 5 10 15 20 25 200 100-(D 50-- 0-50-25 Figure 3-6: Experimental magnitude and phase responses to varying AC current in the opposite direction. The cantilever is driven on resonance. Current levels are 5, 25, 50, 100, 250, 500 pA and 1, 2, 3, 4 mA. The arrows are drawn in the direction of increasing current. Tip-sample distance was 200 nm. 3.3. RESULTS 65 The difference in the degree of asymmetry between the two directions of current indicated possible concurrent sensing of electrostatic force in addition to magnetic force. The Cr on the CoCr MFM tips is conductive and the tips are sensitive to electrostatic force. In fact, the same tips are also widely-used for electrostatic force microscopy (EFM), which measures the voltage difference between the tip and the sample. In our experiments, the tip is grounded while there is a voltage drop through the wire. The EFM magnitude responses for the two current directions are expected to be different. The direction of the voltage drop switches and the potential at the same location along the wire may be different. Fig. 3-7 shows the magnitude responses to electrostatic force alone for the two current directions. Specialized conductive probes from DI were used (Model SCM-PIT). They are Platinum/Iridium coated and are non-magnetic. The phase responses are not shown but they are nearly constant, with only about 2 to 3 degrees of variation. 3.4 x 10 3.2 3 2.4 -2 5 6 -20 -15 -10 -5 -20 -15 -10 -5 0 x {um] 5 10 15 20 25 0 [um] 5 10 15 20 25 x10 5.8- 5.2 5 4-2 5 x Figure 3-7: Magnitude responses to electrostatic force caused by 2 mA of current in both directions. A non-magnetic tip (Model SCM-PIT) was used. CHAPTER 3. AC CURRENT IMAGING 66 3.3.1 Decomposition Method Due to the symmetry of the total EFM response and the asymmetry of the total MFM response, a decomposition method is proposed to separate the total cantilever response into even and odd components, of which the even would represent the EFM response and the odd would represent the MFM response. Fig. 3-8 graphically illustrates the decomposition method. The decomposition involves two operations. The first flips the total response about the z-axis and subtracts it from itself. From Fig. 3-9, it is evident that the odd component is reconstructed by this operation, while the even component is simultaneously cancelled out. The second operation extracts the even component by halving the sum of the flipped response and the original (see Fig. 3-10). 3.3.2 Decomposed Results Figs. 3-11 and 3-12 compare the extracted EFM and MFM response to the experimental magnitude responses. Note that the total MFM response includes both the magnitude and the phase. The MFM responses from the two current directions are identical outside of the 7r phase difference. The EFM responses, however, have different offsets, which agree with previous analysis. The height of the EFM responses also appear to be smaller than that of the MFM. Furthermore, the results from the seven lowest current values from Figs. 3-5 and 3-6 are taken and decomposed to its EFM and MFM responses in Figs. 3-13 and 3-14. Clear difference in the MFM response is seen down to 15 pA, suggesting that the sensitivity in AC current imaging is about 15 pA. 3.3.3 Response to Varying Lift Heights The cantilever response to varying lift heights are shown in Fig. 3-15 and the decomposed EFM and MFM results are showns in Fig. 3-16. As lift height increases, the MFM response broadens and slightly diminishes. The EFM response also decreases. 3.4 Comparison to Theory Having decomposed the cantilever response into an EFM and an MFM component, it is necessary to compare the MFM response to theory. EFM simulations are also compared to the EFM results 3.4. COMPARISON TO THEORY 67 Z Odd Component a -C C X + Z Even Component C -c z (a+b)/2 Total Response a x -c L V (a-b)/2- Figure 3-8: Schematic drawing of the decomposition method. The total response can be represented by the addition of an even and an odd component. 68 CHAPTER 3. AC CURRENT IMAGING Z x z z b/4 _ b/4 C x -c Xx -b/4 z z b/4 b/4_|_ x -c-b/4. -c X . Figure 3-9: Method for extracting the odd (MFM) component. The odd component can be reconstructed by subtracting half of the flipped response from half of itself, while the even component vanishes during the same operation. Therefore, by doing the same to the total response, only the odd component remains. 3.4. COMPARISON TO THEORY 69 z a -C C X Z -C *z C x + a/2 -C z z a/2 -C C C X + a/2 -C C Figure 3-10: Method for extracting the even (EFM) component. The even component can be reconstructed by adding half of the original response and half of the flipped response. The odd component is cancelled out during the same operation. By applying this method to the total response, the even component is extracted. CHAPTER 3. AC CURRENT IMAGING 70 0.035 0.03 0.025 0.02 0.015 0.01 0.005 -20 -10 10 0 20 x [um] x 10' 0.03 15 0.02 5710 0.01 2 .~5 CO 00 -0.01 0 -0.02 -0.03 -20 -10 0 x [um] 10 20 -20 -10 0 10 20 x [um] Figure 3-11: (Top) Total experimental magnitude response due to 2 mA AC current. (Bottom Left) Extracted EFM magnitude response from total. (Bottom Right) Extracted MFM magnitude response from total. 3.4. COMPARISON TO THEORY 71 0.04 0.035 5 003 0.025 0.02 z 0.015 0.01 0.005 -20 -10 0 x[um] 10 20 0.03 0.02 -0.01 CD C: CD co 2 0.01 -0.015 03 0 -0.01 -0.02 -0.02 -0.025 -0.03 -20 -10 0 x [um] 10 20 -20 -10 0 10 20 x [um] Figure 3-12: (Top) Total experimental magnitude response due to 2 mA AC current in the reverse direction. (Bottom Left) Extracted EFM magnitude response from total. (Bottom Right) Extracted MFM magnitude response from total. CHAPTER 3. AC CURRENT IMAGING 72 x x 10 10- 3.5 2.51 _0 :3 'E 0) as 2 '0 1.5 0 0) Ca Z -2 0.5 -4 0 -6 -0.5 -20 -10 0 x [um) 10 20 -20 -10 0 10 20 x [um] Figure 3-13: Experimental EFM and MFM magnitude responses to varying AC current. EFM data is the even component of the total response and MFM data is the odd component. The current levels are 5, 15, 25, 50, 100, 250, and 500 pA. The responses increase with current. Tip-sample distance was 200 nm. 3.4. COMPARISON TO THEORY 73 X 10, x 10 -0.5-1 -1.5-2 :3 0) ca i -2.5 F 'E 0) Cz 2 -3 0 -2 -3.5 F -4 -4 -4.5 F -6 -5 -20 -10 0 x [um] 10 20 -20 -10 0 x [um] 10 20 Figure 3-14: Experimental EFM and MFM magnitude responses to varying reverse AC current. EFM data is the even component of the total response and MFM data is the odd component. The current levels are 5, 15, 25, 50, 100, 250, and 500 pA. The responses increase with current. Tip-sample distance was 200 nm. CHAPTER 3. AC CURRENT IMAGING 74 0.05- 0.02 - S)0.015 -- 0.01 0.005 - -25 -20 -15 -10 -5 0 x [um] 5 10 15 20 25 -20 -15 -10 -5 0 x [um] 5 10 15 20 25 100- 50- 0 -25 Figure 3-15: Experimental magnitude and phase responses to varying lift heights. Current level was at 2 mA, flowing in the reverse direction. Lift height ranges between 200 and 2000 nm and increases in increments of 600 nm. The arrow is drawn in the direction of increasing lift height. 75 3.4. COMPARISON TO THEORY x 10' 0.025-1 0.020.015- -2 0.01 - -3 0.005 a) C-, --4 0 -0.005 -5 -0.01 -6 -0.015 -7 -0.02 -0.025 -8 -20 -10 0 x [um] 10 20 -20 -10 0 10 20 x [um] Figure 3-16: Extracted EFM and MFM magnitude responses to varying lift heights. Current level was at 2 mA, flowing in the reverse direction. Lift height ranges between 200 and 2000 nm and increases in increments of 600 nm. The arrows are drawn in the direction of increasing lift height. CHAPTER 3. AC CURRENT IMAGING 76 0.04 0.030.02 2 .Simulated - 0.01 r q =3.2x1 0~6 Am o - LH =10.5 . um -0.01 q 2.8x10-5 Am LH = 6.5 um -0.02- -0.03- -0.04 -25 -20 -15L -10L -5 0 x [um] 5 10 15 20 25 Figure 3-17: Comparison between the experimental and simulated results of the cantilever magnitude response to a 2 pm-wide wire carrying 2 mA AC current. The current frequency matches the first vibrational resonance of the cantilever beam. The simulation used the extended monopole model to describe the magnetic tip-sample interaction. Two fits were used to match the experiment. The first yields q = 3.2x 10-5 Am and z = 10.5 pm and the second has q = 2.8x 10-5 Am and z = 6.5 pm. The experimental data, however, is in the units of volts and requires conversion into meters, representing the amount of physical tip deflection. This conversion varies with each measurement session because the tip placement and laser alignment differ from session to session. Typically, a series of steps is performed at the end of each session to determine the relationship between the voltage signal and the physical tip deflection. Unfortunately, these steps were not performed when the measurements were made. An estimate of 1 mV to 1 nm is assumed. The process of finding the voltage-to-meter relationship is explained in Chap. 5. 3.4.1 MFM Fig. 3-17 shows the experimental and simulated magnitude responses to 2 mA AC current. The simulations assume the extended monopole model. The first uses the parameter values q = 3.2 x 10-5 Am and z = 10.5 pm and the second uses q = 2.8x10- 5 Am and z = 6.5 pm. The first matches well at further distances away from the wire while the second fits well near the wire. The need for two fits may be because the magnetic field decays faster as a function of 3.4. COMPARISON TO THEORY 77 tip-sample distance near the wire. Fig. 3-18(a) shows the magnetic field strength up to 10 pm away from the center of a 2 pm-wide wire at various tip-sample distances. According to the extended monopole model, the effective displacement of the point pole is dependent on the gradient of the magnetic force in z. If the magnetic field decays at a faster rate along z, then effectively less of the total tip volume interacts with the field; and if the magnetic field decays slowly, then more of the tip interacts with the field and the effective monopole should be modeled as if it were at a higher position along the tip. The large difference in the effective lift height between the two fits appear to agree with the above reasoning. 32- 0 0--1-2 -3-10 -8 -6 -4 -2 0 x [um] 2 4 6 8 10 -8 -6 -4 -2 0 x [um] 2 4 6 8 10 6 420 T-2 - -4-6-10 Figure 3-18: (a) Magnetic field strength and (b) gradient of magnetic field from 2 mA AC current at tip-sample distances ranging from 200 nm to 3200 nm, incrementing every 500 nm. The arrow indicates the direction of increasing lift heights. Furthermore, the effective lift height values used to fit the AC experimental results are significantly larger than those used in the DC case. In AC current imaging, critical information lies in the magnitude response, which is proportional to the magnetic force. This means that AC current imaging is a direct response to the magnetic field because the monopole model asserts a linear relationship between the magnetic force and the magnetic field. On the other hand, information in DC current imaging is given by the phase response, which is a response to the gradient of the magnitude CHAPTER 3. AC CURRENT IMAGING 78 field in z. Fig. 3-18 shows the decay in the magnetic field and gradient of magnetic field as a function of tip-sample distance. Fig. 3-19 shows the maximum Hz and dHz/dz value as a function of tip-sample distance. The height at which Hz and dHz/dz at z = 200 nm decays by l/e are z = 1.2 pm and z = 621 nm respectively. The effective lift heights between the AC and DC simulations differ by an order of magnitude, whereas the placement of 1/e decay between Hz and dHz/dz only differ by a factor of two. Other reasons causing larger effective lift heights in AC current imaging are not understood and remain a question to be answered in future work. 6-%I dH /dz 5-' E o 1/e decay of dH /dz 4- z = 621 nm 1/e decay of Hz z = 1.20 um 3 XI ................................. ............... ........ 2- H 01 0.2 0.4 0.6 0.8 1.2 1 1.4 1.6 1.8 2 z[um] Figure 3-19: Maximum Hz and dHz/dz as a function of z. Hz at z pm, and dHz/dz decays 1/e at z = 621 nm. = 200 nm decays l/e at z = 1.2 Fig. 3-20 shows comparison between the experimental and simulated MFM responses to 2 mA AC current at various tip-sample distances. q is 3.2x10-- Am. The physical lift heights d range from 200 nm to 2000 nm. The fitted effective lift heights values, z, increase at a slower rate and range from 10.5 pm to 11.6 pm. As the physical tip-sample distance increases, the 1/e decay rate decreases. Therefore, the effective lift heights do not increase linearly with the physcial lift heights. 3.4. COMPARISON TO THEORY 0.02 79 d = 200 nm z=10.5um 0.021 0.01 0.01 -E ) 0 /- ) -0.01 -0.02 -20 5;. -a) 0 -0.01 /- -0.02 0.02 d = 800 nm z=10.8um -10 0 x [um] d = 1400 nm z = 11.2 um 10 20 -20 0.02 / 0.01 57 0 -0.01 0 x [um] 10 20 20 L . =2 /- -0.02 -10 10 / /- -20 20 ' 0) -0.02 10 0.01 -a) -0.01 0 x [um] d = 2000 nm z=11.6 um 0 /- ca) -10 -20 -10 0 x [um] Figure 3-20: Comparisons between experimental and simulated MFM responses at various lift heights. The solid lines represent experimental results and the dashed lines represent simulated results. CHAPTER 3. AC CURRENT IMAGING 80 3.4.2 EFM EFM simulations are also needed to check the experimental results. Electrostatic force is caused by a difference in voltage between the tip and the sample. And since force is the derivative of energy, electrostatic force can be expressed as a function of the difference voltage, V, and its capacitance, C[1],[11]: V2 C Fz (3.7) 2 Oz A parallel-plate capacitor is assumed, with an effective surface area of V/R where R = 40 nm[11]. The tip-sample distance represents the distance between the parallel plates. Furthermore, fringing effects are neglected and F, is only experienced when the tip is above the wire. Fig. 3-21 shows the simulated EFM response to 2 mA AC current. Because the simulation assumes no electrostatic force outside the region of the wire, no magnitude nor phase response is detected. The symmetry of the EFM response agrees with the experimental results. However, the magnitude response is nearly three orders of magnitude smaller than that from the experiments. Presently, the reason is still unknown. x10, E 1.510.5- -25 -20 -15 -10 -5 0 x [um] 5 10 15 20 2! 5 -20 -15 -10 -5 0 x [um] 5 10 15 20 25 1D0 80604020-25 Figure 3-21: Simulated EFM response to 2 mA AC current. V = 100 mV. 3.5. SUMMA RY 3.5 81 Summary The chapter establishes theory for AC current imaging by using new boundary conditions to solve the flexural beam model. Experimental results are shown and confirm the theory. Four key points are concluded from the discussion of AC current imaging. First, from the theory and simulations, the response to AC current is contained in the cantilever magnitude response. Therefore, AC current imaging is a direct response to the magnetic field and not its gradient. However, the magnetic field as a function of tip-sample distance decays slower than the gradient of the field and requires the use of a larger effective lift height. The effective lift height used to fit the experimental response to 2 mA AC current at a physical lift height of 200 nm is 10.5 Pm, with an effective monopole moment of 3.2x 10-5 Am. Therefore, even though the sensitivity of AC current imaging is around 15 pA, the large effective lift height will reduce the original spatial resolution of MFM. Lastly, we provided a decomposition method which separates the MFM and EFM responses from the experimental data, upon recognizing that the MFM response corresponds to the odd component of the total response and the EFM response correscopnd to the even. 82 CHAPTER 3. AC CURRENT IMAGING Chapter 4 Method for Non-Linear Tip-Sample Interaction The presence of both a piezo and an AC current drive causes non-linear tip-sample interactions and requires a new method to describe the cantilever response. Several authors have examined the tip-sample interaction caused by the periodic contact made on the sample surface by the tip in tapping mode and reported higher-harmonic generation by non-linear interaction[15],[36]. Stark et al. also used Fourier transform principles to mathematically reconstruct the tip-sample interaction from the spectral response[37]. However, no general mathematical model is currently available. This chapter presents the non-linear theory. The limiting case where the current frequency approaches zero is examined and used to validate the theory by comparing the results to DC current imaging responses. By reducing the piezo drive amplitude to zero, the theory is also compared to AC current imaging. One direct application of the non-linear theory is in the area of parametric gain. The AC current source at frequency w, transfers the power applied to the piezo at frequency wp to a new frequency, W, + WP. By matching the sum of the two frequencies to a cantilever vibration mode, the response is enhanced by the high Q of the cantilever mode. This enables the piezo to be driven much harder because it can now be driven off resonance and rely on the current frequency to transfer power to the resonance. At the end of the chapter, the intrinsic non-linearity in the AFM system is examined. 83 CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 84 4.1 Theory If the piezo is driven at frequency wp and with amplitude a and the current has frequency wc, then the new boundary conditions are as follows: (4.1) z(x = 0, t) = a cos wpt z(x-0,t) = 0 (4.2) El 0 z(x=L,t) = 0(4.3) z~x=L-Lt) EI EIz where Fz (zo) and - x-L,t) = Fz(zo) coswct + ( Ijzo coswct)z(x = L, t) (4.4) Izo are the first two Taylor series components of a DC magnetic force since the time-varying terms are included externally, and zo is the effective tip-sample distance. From Eqn. 4.4 we see that the first Taylor series term drives the cantilever at the current frequency and that the second term contributes to the generation of higher-order harmonics and intermodulation products. Consequently, a general solution for z(x, t) can be guessed, assuming the higher-orders are sufficiently smaller: z(x,t) = Zdc(x) + Zca(x)eJwct + Zcb(x)e-3wct + ZCc(x)eJ2wct + Zcd(x)e- 2_2wc t + Zpa(x)e31pt + Zpb(x)e-wPt + Zcpa(x)e3(w±+wP)t + Zepb (x)e3(w -w )t + Zcpc(x)eJ(-wc+wp)t + Zcpd(x)e-3(wc+wp)t (4.5) 4.1. THEORY 85 where (4.6) Zdc(x) = dc1 +dc2 x+dc3 x 2 +dc 4x 3 Zca (x) = ca 1 (cos kcax + cosh kcax) + ca 2 (cos kcaX - cosh kcax) + ca3(sin kcax + sinh kcaX) + ca4(sin kcaX - sinh kcaX) Zeb (x) Zcpd(x) = cb1 (cos kebx + cosh kCbx) + cb2 (cos kebx - cosh kcbx) + cb 3 (sin kcbx + sinh kebx) + cb4 (sin kcbx - sinh kcbx) (4.7) cpd1(cos kcpdx + cosh kcpdx) + cpd2 (cos kepdx - cosh kepdx) + cpd3 (sin kcpdx + sinh kcpdx) + cpd4 (sin kcpdx - sinh kcpdx) The spatial distribution of each frequency component contains four coefficients that need to be determined by the boundary conditions. must satisfy Zdc(x), because it is not related to a time-dependent term, = 0 (see Eqn. 2.26), of which Eqn. 4.6 is a general solution. Each of the ten d 4d" other spatial distributions (Zca(X), Zcb(x), etc.) require a dispersion relation relating its frequency component to its wave number. The method is identical to the situation where only one frequency component is present (Sec. 2.1.3), because each frequency component is independent from another. For example, to find the dispersion relation between wc and kca, Za(x)elt is plugged into the damped beam equation (Eqn. 2.26), and the relationship is found: ca= (2 4Q 2J ) (4.8) Or for Zcpa(x)e3(oc+WP)t, (PC+ W) where Wca = k2ca'a =W ( 4 1- +1 ) (4.9) E = k2 L and wcpa pa cpa VpA~ pA The same relationship applies for the terms that have positive frequency components. In the case of a negative frequency component, such as Zc6 (x)e3wc , the dispersion relation is slightly different, WC where Wcb = k V . ( 4 - - ) (4.10) CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 86 It is important to note that if w, > wp, then (-we + wp) is negative and (w, - wp) is positive. Likewise, if wp > we, then the signs are flipped. Once the dispersion relations are established, the coefficients can be determined using the boundary conditions. The first three boundary conditions do not generate any new frequencies and the 11 spatial distributions remain independent from each other. By applying Eqn. 4.1 to the general equation, the first coefficient of each spatial distribution is zero except for pal and pbi, which equal a/2. And from Eqn. 4.2, dc 2 and the third coefficients of the others (i.e. ca 3 , cb 3) are also set to zero. Eqn. 4.3 relates dc3 and dc4 and the second and fourth coefficients of the others. The last boundary condition relates the spatial distributions to each other. By collecting the terms with the same frequency dependence together, 11 new equations emerge: 3 DC wc -wc 2w, d Zd (x =L) = 3 dx d3 Ze. x=zL) =3 3 3 3 d Z dx(x=L) -= d Z,dx(x=L) M1Zeb(x = L) + M1Zca(x = L) +± M1Zdc(x = L) + M 1 Zee(x MlZcd(x = L) + MlZdC(x = L) + - L) M1Zca(x = L) 3 -2wc wp - wp (w + wp) (w - wp) Swp) (-we d Zcd(x=L) 3 d33Z, dX(x=L) = d Z ,(x=L) 3 dx 3 d Z b(x=L) = 3 dx d3 Z~~p.(x=L)- M1Zep (x MlZepd(x MlZeb(x L) L) = L) + M 1 Zcpa (x = L) + M1Zcpb(x = L) MlZpa(x = L) 3 dx d3 Z,,b(x=L)- MlZpb(x = L) M1Zpa(x = L) 3 dx d Z~pd(x=L)3 dx 3 - (we where Mo = Swp) F (zo) and M, = 1 MlZpb(x = L) K Together, the 22 equations from the third and the fourth boundary conditions are enough to solve the 22 unknown coefficients, two from each spatial distribution. A 22x22 matrix scheme is 4.1. THEORY 87 used and each row represents one equation, A1,1 A, 2 A1,3 A 2 ,1 A 2 ,2 A 3 ,1 A 3 ,2 A 1 ,4 A 2 ,3 ... ... ... ... ... ... ... A 2 2 ,1 ... ... A 4 ,1 A ... 1 , 22 dc3 B1 ... dc4 B . .. ca 2 ... ca4 ... cb2 B5 ... cpd 4 B22 2 B 3 = B 4 (4.11) CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 88 where A A A A A A , 1 2 6L T(kca) A , 2 4 U(kca) ,5 T(kcb) A 3, 6 U(kcb) 4 ,7 T(kcC) A 5 ,9 T(kcd) A T(kpa) A 1 ,1 2 ,3 3 A 6, 11 A A A A A A : 7 ,1 3 T(kpb) A 8 ,1 5 T(kcpa) A 9 ,1 T(kcpb) A 7 4 ,8: U(kCC) 5 ,1 0 U(kcd) 6 ,1 2 U(kpa) 7 ,1 4 U(kpb) 8 ,1 6 U(kcpa) 9 ,1 8 U(kcpb) 10 ,1 9 T(kcpc) A 1 0 ,2 0 U(kcpc) 11 ,2 T(kcpd) A 1 1 ,2 2 U(kepd) 12 ,2 -6 A 1 2 ,3 MiX(kca) A 1 3 ,2 Ml L A 1 3 ,8 1 A1 , 2 6 A 13 ,1 13 ,7 14 ,1 MI L M 1 X(kcc) A A A ,4 : MiY(kca) A 12 ,5 : M1X(kcb) A 13 ,3 : kaV cca) A 13 ,4 : kgcW(kca) A 14 ,5 : k3bV(kcb) A 14 ,6 kb W(kc k3cV(kcc) A 15 ,8 k cW(kcc) IcdV(kcd) A 16 ,1 0 cdW(kcd) 17 ,1 6 MI Y(kcpa) 18 ,1 8 MY(kcpb) 19 ,1 6 WpaW(kcpa) 20 ,1 8 k cpb W(kcpb) 21 ,2 0 22 ,2 2 MY(kcc) 2 A 12 3 2 A A MI Y(kcb) M 1 L 3 A A 14 ,9 MX(kcd) 15 ,3 MiX(kca) A 16 ,5 Ml X(kcb) A A MI L 14 ,2 14 ,1 0 MY(kcd) 15 ,4 MiY(kca) A 16 ,6 M 1 Y(kcb) A 17 ,1 2 k3aW(kpa) A 18 ,1 4 IebW(kpb) A ,1 2 MaY(kpa) 19 A 20 ,1 4 M A 1 7 ,1 1 A 1 8, 13 k3bV(kpb) A 1 9 ,1 1 M 1 X(kpa) A A 2 0, 13 M1X(kpb) A A 2 1, 11 M 1 X(kpa) A 2 1 ,1 2 A 2 2, 13 M A 2 2 ,1 4 1 X(kpb) A B B B B Y(kpb) A MiY(kpa) A M 1 Y(kpb) A 1 -*X(kpa) 6 16 ,9 17 ,1 5 M 1 X(kcpa) A 18 ,1 7 MiX(kcpb) A 19 ,1 5 k paV(kcpa) A 20 ,1 7 k PbV(kcpb) A 21 ,1 9 22 ,2 B B 13 4kpaS(kpa) 17 19 B21 1 5 ,7 : -M M 1 W(kpa) 1 W(kpa) B B B A kpcV(kcpc) k3 1 V(kcpd) pb) 7 14 a k3bS(kpb) 18 20 22 : - TM1 W(kpb) - M1W(kpb) A k 6 ) W(kp) W kp cpd W(kcpd) 4.2. APPLICATION TO PARAMETRIC GAIN 89 and all the other elements are zero. A key to the abstractions are as follows: S(k) = sin kL + sinh kL T(k) = -cos kL - cosh kL U(k) - sin kL - sinh kL V(k) = -sinkL + sinhkL W(k) = cos kL + coshkL X(k) = cos kL - cosh kL Y(k) = sin kL - sinh kL The coefficients are solved numerically in Matlab by multiplying the inverse of matrix A by matrix B. Once the coefficients are found, the response at difference frequencies can be determined. For example, the complete solution at w, is Z,(x) where IZc(x)I = 21Za(x)I = 2IZeb(x)I and LZ LZa(X) 4.1.1 = = -LZZb(x). Limit to DC and AC Current Imaging In order to validate the non-linear theory, the cantilever response at the limit where the current frequency approaches zero is compared to the DC current response shown in Chap. 2. The piezo is driven on resonance. The result is shown is Fig. 4-1, and it is identical to Fig. 2-8. At the limit where the piezo drive amplitude is zero, the cantilever is again purely responsive to the AC current and the result is identical to AC current imaging (compare Figs. 4-2 and 3-1). 4.2 Application to Parametric Gain The non-linear system, including both piezo and AC current drives, provides new possibilities for cantilever excitation. One is to examine the w, + w frequency component. By matching w, + wp to the cantilever's first resonant frequency, the power applied to the piezo drive can be tranferred to the resonance by the current drive. These principles of parametric gain are first discussed. The degenerate case where W, = WP is presented next, followed by the nondegenerate case where the two frequencies are different. CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 90 2820 28002780 R 276002740272027 -10 -8 -6 -4 -2 0 x [um] 2 4 6 8 10 -8 -6 -4 -2 0 x [um) 2 4 6 8 10 10 0 .1 - 10 -10 Figure 4-1: Magnitude and phase response to 20 mA current at the limit where w, approaches zero. WP is driven on resonance . q = 10-6Am. z = 400 nm. 4.2.1 Principles of Parametric Gain Parametric gain can be illustrated by a damped harmonic oscillator in which an energy storage parameter is modulated at some frequency wi [40]. An example of a degenerate parametric oscillator is an RLC circuit in which the capacitor is time-varying. Consider a normal parallel RLC circuit. The voltage across the elements as a function of time is d2 v(t) 1 dv(t) dt 2 d2RC + dt 1 LC v) t = =0 (4.12) However, if the capacitor has an added time-varying term, where C = Co + ACsinwit (4.13) a new expression for the voltage would be d2 v(t) 1 dv(t) 1 AC dt2 + RC0 dt + LC0 (1± C0 sinwitv(t))= (4.14) 4.2. APPLICATION TO PARAMETRIC GAIN 91 70 605040 0 'E20 -20 -- 10 -20 -15 -10 -5 0 x [um] 5 10 20 15 -0) 100 50- 0-- -50- -100 -20 -15 -10 0 -5 5 10 15 20 Figure 4-2: Cantilever response with no piezo drive to 2 mA AC current flowing through a 2pm-wide wire, centered at x = 0. q = 8 x 10-6Am. z = 1 pm. assuming that AC < Co. By assuming a solution v(t) = a cos(wt + #) and neglecting the (wi + w) term, Eqn. 4.14 can be solved: (P _ edot+0) + RC RCo (2)e3(-t++) where wo = LCO Therefore, if w1/2 = w = wo, # = 0 or -r, and RAC 2Co = 0 =-e3[W)t*] (4.15) 2/wo, then steady-state oscillation can be achieved. Physically, the time-varying capacitance can be thought of as a parallel-plate capacitor whose distance between the two plates varies with time, since capacitance is inversely proportional to distance. Fig. 4-3 schematically shows the voltage gain achieved by the modulated capacitance. Because the capacitance changes at twice the resonant frequency, if the plates are pulled apart at the maxima and minima of the charge function, q(t) and are pushed together at the zeros of the charge function, parametric gain in the voltage function can be achieved. At the instance right before the plates are pulled apart, a certain amount of charge accumulates on the plates (Fig. 43(a)). Once pulled apart, the capacitance drops but the charge cannot change instantaneously and a) t = ti b)t= t2 d, d2 = d1 +8 Figure 4-3: Schematic drawing of time-varying parallel-plate capacitor. (a) At t = ti, charge accumulates on the parallel-plates. q(ti) = C(ti)v(ti). (b) At t = t 2 , the parallel plates are pulled apart some distance 6. Instantaneously, the charge remains the same, q(t 2 ) = q(ti), while the capacitance drops, C(t 2 ) < C(ti). Therefore, because q(t 2 ) = C(t 2 )v(t 2 ), v(t 2 ) > v(t)1. remains the same. Therefore, to satisfy q(t 2 ) = C(t 2 )v(t 2 ), the voltage increases instantaneously. In other words, the work done to pull the parallel plates apart was given back as voltage gain. On the other hand, no work is needed to push the plates back since the timing coincides with the zeros of the charge function. Another example of a damped harmonic oscillator is the mass-spring system, from which the point-mass model is derived. By relating the elements of a mass-spring system to the RLC circuit, intuition about the non-linear cantilever response can be gained. The effective spring constant k, of the cantilever beam in the point-mass model is analogous to the capacitor. Since magnetic force on the tip shifts the effective spring constant, k, is modulated at the AC current frequency w. The force on the effective spring and its displacement is related by F(t) = k(t)z(t) where F(t) is analogous to q(t) and z(t) is analogous to v(t). If the maxima and minima of the force function coincides with the "loosening" of the spring, then an instantaneous gain in displacement must occur to satisfy the force equation. 4.2.2 Degenerate: w, = wp Parametric gain in the degenerate case can be shown in the response of 2W frequency component. With w, = wp = W, the response of the eJ2wt term is maximized when 2w matches the resonant frequency of the cantilever. Fig. 4-4 shows the magnitude responses of both the e-wt and the e 2 wt terms as a function of drive frequency. The frequencies at which the peaks occur are evident: for eswt, it happens when w = wo; for e 2"t, it happens when w/2 = wo. 4.2. APPLICATION TO PARAMETRIC GAIN 93 3000 250020001500100050040 50 60 70 80 Frequency [kHz] 90 100 110 50 60 70 80 Frequency [kHz] 90 100 110 0.014 0.012 0.01 0.008 8 0.006 2 0.004 0.002 40 Figure 4-4: Magnitude response to 2 mA AC current through a 2pum-wide wire, centered at x = 0 across sweep of frequencies. w, = w= ,. q = 5x10-7Am. z = 800nm. (Top) Response of e3" term. (Bottom) Response of eJ2wt term. 4.2.3 Nondegenerate: we wp Fig. 4-5 shows the magnitude responses to 1 mA AC current of various combinations of W, and wp, with the criteria that w, + wp equals the first vibrational frequency, wo. The piezo is not driven in the top figure, and w, = wo = 89.9 kHz. This is identical to AC current imaging and the maximum response is 42 nm. In the middle figure, w, = 34.9 kHz and wp = 55 kHz. The piezo drive amplitude required to achieve the same cantilever response as the top figure is 12 pm. The piezo frequency in the bottom figure is 100 Hz less than the resonant frequency and the drive amplitude needed is only 8 nm. By driving the piezo near the resonant frequency, the response is enhanced drastically by the of the cantilever. Fig. 4-6 compares the cantilever response at wp = wo Q - 187 Hz and wp = wo - 100 Hz. w, = 0 and 100 Hz respectively. The first corresponds to DC current imaging at the frequency of optimal magnitude response. The slope of cantilever response versus piezo drive amplitude is the same for the two frequency settings as well as for different current levels. However, the DC current response is more than two CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 94 40- E £30. 20 10 -15 -10 -5 0 x [um] 5 40 - a E 0) 20 CU -15 10 15 10 15 1.5 um a =1.2 um -10 -5 0 x [um] 5 60- - = 12 nm 4-a S40 o20 01 -15 a=8 nm -10 -5 0 x [um] 5 10 15 10-5 89.9 kHz. q = Figure 4-5: Magnitude responses to 1 mA AC current when w, + w, Am. z = 1 pm. (Top) w, = w0 and w, = 0. (Middle) w, = 34.9 kHz and wp = 55 kHz. The piezo drive amplitude, a, needed to be about 12 pm to match the magnitude response of the top figure. (Bottom) w, = 100 Hz and w, = 89.8 kHz. The piezo drive amplitude only needed to be 8 nm to match the magnitude. orders of magnitude larger. The simulation uses fitting parameter values similar to those determined in DC current imaging (q = 10- Am and z = 1 pm), and they are assumed to be the same for both frequency settings. But without nondegenerate experimental data with which to compare, this assumption cannot be verified. Therefore, focus should be on the sensitivity to current level within the two frequency settings. When wp = wo - 100 Hz, the change in cantilever response to current level is very drastic. Fig. 4-7 zooms in on the DC current response to examine its change to current level. This concludes that the nondegenerate case is more sensitive to current level. However, two practical issues remain. First, q and z must be fitted to experimental results in order to accurately simulate the physical cantilever response. Second, the noise floor must be examined to determine a feasible sensitivity level. 95 4.2. APPLICATION TO PARAMETRIC GAIN 106 ........ lo 102 1 nA, 1, 2, 10 mA S100 10 mA 2 mA - Nondegenerate ..... DC current - 1 mA S10~ -) 1 nA 10-8 10' 10 Piezo drive amplitude [nm] 10 102 Figure 4-6: DC current and nondegenerate cantilever response to varying piezo drive amplitude. DC current: wp = wo - 187 Hz and w, = 0. Nondegenerate: w, = wo - 100 Hz and w, = 100 Hz. q = 10- 7 Am and z = 1 ptm. x010 454.I 7 4.5654.56 4.555 4.55 mA -10 . 4.545 4.54 2 mA 0 4.535 - 4.53 1 mA 4.525 I nA 4.52 200 200.2 200.4 200.6 200.8 201 201.2 Piezo drive amplitude [nm] 201.4 201.6 201.8 Figure 4-7: DC current cantilever response at w, = wo - 187 Hz and w, amplitude. q = 10-7 Am and z = 1 pm. = 202 0 to varying piezo drive CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 96 4.3 Non-Linearity in AFM System The nonlinear response of the AFM system is examined to ascertain that the non-linearities observed experimentally are mainly from the non-linear tip-sample interaction and not from the system itself. The test for non-linearity in the AFM system was performed with a regular tapping-mode tip. It was driven by the piezo and kept far above the sample to ensure no interaction. The piezo drive amplitude was increased slowly and both the input piezo voltage and the output voltage from the photodetector are recorded by the oscilloscope and Fourier transformed into the frequency domain. One raw FFT spectra of output voltage is shown in Fig. 4-8. -20- -40- E -60 C. -80 -100- -120- -140' 0 500 1000 1500 2500 2000 3000 Frequency [kHz] 3500 400 4500 5000 Figure 4-8: FFT spectra of output voltage to 0.205 input voltage. The input and output voltages at the drive frequency as well as two and three times the frequency are plotted against the input voltage at the first harmonic. The figures are shown in Figs. 4-9 and 4-10. The plot of vi, at the first three harmonics versus vi, at the first harmonic is used for reference. The slopes of the higher harmonics are all much smaller, suggesting minimal generation of nonlinearities in the system. The slope of the output voltage at the first harmonic is very close to one and confirms its response to the input voltage. By examining the difference among the harmonics at vi, = 0.257 V, non-linear generation can be determined by comparing the results of the input and the output voltages. The difference between 4.4. SUMMARY 97 the first and second harmonic input voltages is 0.233 V, whereas the same comparison of the output voltages yields a difference of 0.232 V. This shows that there is no generation of the second harmonic between the piezo drive and the photodetector output. The same comparison between the first and third harmonics yield 0.248 V for the input voltages and 0.238 for the output voltages. Therefore, no third-harmonic-generation is observed either. This concludes that the non-linearity in the AFM system is minimal. 4.4 Summary In this chapter, a general method for non-linear tip-sample interaction is presented and shown to be consistent with DC and AC current imaging theories in Chaps. 3 and 4. By using the sum of the current and piezo frequencies to excite the first vibrational mode, parametric gain is achieved. The degenerate case shows a large cantilever response at wo/2 in the 2w frequency component when w, =P = W= wo/2. The nondegenerate cantilever response is enhanced when the piezo is driven near the resonant frequency. The results are compared to DC and AC current imaging. The difference in the nondegenerate responses to varying current level is much more drastic than the DC current responses. And by an 8 nm piezo drive amplitude at wp = wo - 100 Hz, nondegenerate response achieves the same cantilever oscillation amplitude as the pure AC current response. The comparisons to DC and AC current imaging assumed the same effective monopole moment and displacement values respectively. However, in order to accurately determine these values, experimental results are needed. Lastly, the intrinsic non-linearity in the AFM system is determined to be negligible. 98 CHAPTER 4. METHOD FOR NON-LINEAR TIP-SAMPLE INTERACTION 0.4 1 1 1 Fundamental mode 0.35- slope = 1 0.3 - 0.25- - - 0.2 - 0.15Second harmonic 0.05 - Third harmonic slope = 0.028 slope = 0.122 0.1- - - 0 0.2 0.22 0.24 0.26 0.28 0.3 Vi [V] 0.32 0.34 0.36 0.38 0.4 Figure 4-9: Non-linearity in input piezo voltage at the first three harmonics. 0.4 Fundamental mode 0.35- slope = 0.957 0.3 - 0.25 - 3 - 0.2- Third harmonic slope = 0.068 0.15Second harmonic slope = 0.105 0.1 0.05 - 00.2 0.22 0.24 0.26 0.28 0.3 Vin IV 0.32 0.34 0.36 0.38 0.4 Figure 4-10: Non-linearity in output voltage of a free cantilever at the first three harmonics. Chapter 5 Conclusion 5.1 Summary The goal of this thesis is to explain DC and AC current imaging in magnetic force microscopy as well as understand the effects and applications of non-linear tip-sample interactions. Chap. 2 derives the flexural beam model and presents various methods to describe the magnetic tip-sample interaction. DC current imaging theory is subsequently presented and compared to the experimental data. The extended monopole model matched the experimental results better than the extended dipole model. The effective monopole moment and displacement for 15 mA DC current were found to be 1.9 X10-7 Am and 850 nm respectively. Magnetic imaging at the second cantilever vibration mode was also examined with a standard magnetic tape. Even though the Q is higher at the second mode, the sensitivity decreases because of a faster increase in the effective spring constant. The overall sensitivity to DC current imaging was determined to be at 1 mA. AC current imaging is discussed in Chap. 3. By filtering the response at the current frequency, the results were independent of the piezo drive frequency. Simulations show that the critical magnetic field information is contained in the cantilever's magnitude response, suggesting that the tip responds directly to the magnetic field in AC current imaging. Electrostatic force contamination in the overall cantilever response necessitated a decomposition method which separates the total response to its even and odd components, of which the even represented the tip's response to electrostatic force and the odd represented its response to magnetic force. Comparison with the experimental MFM results suggested a need to employ a much higher effective lift height in the AC simulations (q = 3.2x 10-5 99 CHAPTER 5. CONCLUSION 100 Am and z = 10.5 pm for 2 mA AC current). Even though the current sensitivity was determined to be around 15 pA, the high effective tip-sample distance lowers the spatial resolution of this method. Chap. 4 presents a general theory for non-linear tip-sample interaction, which is not limited to current measurement nor MFM. Comparisons to DC and AC current imaging validate the theory. Parametric gain with the non-linear theory is shown in the degenerate and nondegenerate cases. The nondegenerate cantilever response is enhanced when the piezo is driven near the resonant frequency. With the sum of the piezo and current frequencies set to the first vibrational frequency and L = 100 Hz, the nondegenerate responses show greater current sensitivity than DC current imaging. Lastly, the non-linearity in the AFM system was experimentally determined to be negligible. 5.2 Directions for Future Work Suggestive directions in modeling and experimental work for current measurement research in MFM are presented below. The list focuses on possible next steps in the research. 5.2.1 Modeling Magnetic Tip-Sample Interaction Theoretical development in this thesis focused heavily on the transverse vibration of the cantilever beam due to interest in the higher-order vibrational modes. Less emphasis was placed on the modeling of the magnetic tip-sample interaction. While the use of the extended monopole model is verified experimentally, it remains an approximation. To model the magnetic tip-sample interaction more precisely, the Fourier transform method can be employed[16]. It uses transfer functions in the Fourier domain to simplify the convolution of the magnetic field from the sample and the magnetic surface area of the tip. According to McVitie et al.[22], who evaluated the point charge and Fourier transform models against measured magnetic field distributions from MFM tips, the latter provided the best match. Large effective lift height in AC current imaging The need for large effective lift heights ( 10 pm) in AC current imaging remains a mystery. The original hypothesis attributed this need to the slower decay rate of the magnetic field versus the gradient of the field. However, upon comparing the two l/e decay length at a tip-sample distance 5.2. DIRECTIONS FOR FUTURE WORK 101 of 200 nm, the result for the magnetic field was only a factor of two larger than that of the gradient. According experimental results obtained by Lohau et al.[19], the effective displacement appear to relate linearly with the decay length. Therefore, it does not fully explain the near factor of 10 increase in the effective lift heights between DC and AC current imaging. EFM More detailed EFM simulations are also needed. The current model approximates the capacitance between the MFM tip and the sample as a parallel-plate capacitor, with an effective area of a circle with radius R = 40 nm[11]. However, the magnitude of the simulated EFM response is nearly three orders of magnitude smaller than the experimental EFM results. Noise analysis A more comprehensive discussion of current sensitivity would require detailed analysis of the possible sources of noise. Noise floors for the different imaging methods should be determined. Butt et al.[5] examined thermal noise in the first six vibrational modes. Salapaka et al.[33] presented an overall analysis on multi-mode noise. 5.2.2 Experimental MFM with shielded sample To ensure pure MFM response without EFM contamination, the wire sample can be shielded with a layer of PMMA, which serves as insulation. Therefore, the voltage across the wire sample is shielded from the tip and electrostatic force should be eliminated. EFM with demagnetized MFM tips If AC current imaging is performed without a shield sample, more experimental EFM results are needed. In Chap. 3, the decomposed EFM response from a MFM tip is compared to the response of a non-magnetic EFM tip. The two results are similar qualitatively. However, quantitative comparisons could not be made because the physical dimensions and makeup of the tips are different. In order to ascertain the decomposed EFM response quantitatively, one possibility is to perform EFM with a demagnetized MFM tip. Demagnetization of MFM tips can be achieved by heating the tip for several hours or by allowing the tip to undergo AC demagnetization. CHAPTER 5. CONCLUSION 102 AC current imaging at second harmonic Second harmonic imaging of AC current should also be explored. The magnitude response does not have the same dependence on Q and k, as the phase response does in DC current imaging and better sensitivity may be achieved at the second vibration mode in AC current imaging. Parametric gain Chap. 4 shows the theoretical potential for parametric gain with piezo and current drive frequencies. Experimental verification is needed. Furthermore, q and z must be fitted to experimental data. The phase between the piezo and current drive frequencies must remain linearly related and the lock-in should reference the sum frequency. Conversion from electrical amplitude to mechanical displacement The cantilever oscillation is measured by the photodetection system, which outputs a voltage signal. In order to determine the cantilever oscillation in nanometers, an added step is needed at the end of each experimental session. The method to measure cantilever oscillation uses the TappingMode force curve function in the DI instrument menu. The force curve graphs cantilever oscillation amplitude against vertical piezo scanner movement (see Fig. 5-1). The scanner begins on the right side of the graph, where the cantilever vibrates in free air. As the scanner lowers and the tip begins to tap on the sample surface, the feedback system reduces cantilever oscillation amplitude, which is represented by the downward slope on the force curve. Eventually, as the scanner continues its downward movement, the tip hits the surface and its amplitude reduces to zero. Therefore, the horizontal length of the downward slope represents the peak cantilever amplitude and can be compared to the peak voltage signal. 5.3 Final Words The tradeoff between current sensitivity and spatial resolution between DC and AC current imaging became apparent. While MFM is able to sense AC current down to 15 pA, the large effective lift height suggests diminished spatial resolution. Its performance appears to be more comparable to scanning hall probe microscopy, which has about 1 pA current sensitivity and 1 pm spatial resolution. The non-linear approach to current measurement is promising. Parametric gain can be achieved 5.3. Cantilever amplitude Cantilever oscillation amplitude [nm] 0 Vertical scanner movement [nm] Figure 5-1: Force calibration plot for voltage to metric conversion. and sensitivity to current is enhanced. However, experimental work is needed to support the theoretical analysis and to determine its spatial resolution. The development of current measurement in MFM can serve as a powerful tool for various research. It may provide an simple and accurate way of quality assurance on integrated circuits. Moreover, it allows imaging of current flow through various materials and insights on electron transport can be gained. 104 CHAPTER 5. CONCLUSION Appendix A Experimental Setup This appendix covers the parts of the experimental setup that is common in both DC and AC current imaging. The basic setup of the instruments and the fabrication processes of the wire sample are first presented, followed by details of the data acquisition and process software. A set of experimental steps involved in DC and AC current imaging are also provided. Familiarity with MFM is assumed. A.1 A.1.1 Instrument Setup Atomic Force Microscope The usage of the AFM follows standard D13000 MFM procedures, in terms of the tip calibration, laser alignment, and sample loading processes. A scan rate of 0.1 Hz is used to ensure detailed imaging as well as clean data acquisition by the lock-in. Typically, the lift height during LiftMode is set at 200 nm because the sample has a height of 200 nm. A.1.2 Lock-in Amplifier The values used for the lock-in amplifier are listed in Table A.1. asterisk (*) are controllable through LabVIEW user interface. 105 The controls marked with an APPENDIX A. EXPERIMENTAL SETUP 106 Instrument Control Value GPIB address* 8 Sensitivity* 300 mV (DC) 100 mV (AC) Time Constant* 30 ms Filter 24 dB/Octave Channel 1 R [V] Channel 2 Reference in External, 50 Q Signal in 1MQ, 30 pF Table A.1: Experimental settings for the lock-in amplifier. A.2. SAMPLE FABRICATION A.2 107 Sample Fabrication The wire sample was prepared by Mathew Abraham. A brief description of the process is included here. First, a 100 nm layer of oxide was grown on a 4-inch Si wafe in an oxidation tube, in order to electrically isolate the metal wires and the Si substrate. Al wires were then defined on the oxide layer by a liftoff process. AZ5214 was used as the image reversal resist and 200 nm of Al was deposited using an E-beam evaporator. A.3 Data Acquisition and Processing Software Data acquisition of the lock-in amplifier is controlled through LabVIEW. Once the values are stored into a computer file, Matlab code is used to process and plot the data. A.3.1 LabVIEW Code Due to the graphical nature of LabVIEW programming, it is difficult to include the actual code. Therefore, only the basic functions of Controller.vi is described. First, the data buffer of the lock-in is cleared. Then the instrument parameters are set, including time constant, sensitivity, and sampling rate. Once data sampling is triggered, the buffer begins to store data. The program waits long enough to scan in a specified number of scan lines and queries the stored data, which is subsequently placed in files named by the user. A.3.2 Matlab Code lvread.m reads the stored values from the file specified by LabVIEW and translates them into a matrix in Matlab. A.4 Recipe for DC and AC Current Imaging The important steps in DC and AC current imaging are listed. General knowledge of MFM imaging with the Dimension 3000 is assumed. First, the SAM needs to be connected to the AFM, between the extender electronics module and the microscope. The controller should be turned off and the user should exit the DI program prior to adding the SAM. The tip is then loaded onto the holder and magnetized in the z-direction. APPENDIX A. EXPERIMENTAL SETUP 108 Standard laser alignment and frequency sweep routine are performed and the interleave parameters are set. The tip is calibrated with a magnetic tape and its magnetization is checked by phase imaging in the interleave mode. Typical parameters are as follows: the scan size is 3 to 5 pm; the lift height is 40 nm; the phase z-range is about 30±10 degrees; the height z-range is 100 nm. Note that for better tracking purposes, the tip should engage with the scan size set to 1 prm and increased to the desired scan size after it is engaged. Once the tip's topographic and magnetic force imaging abilities are proven, the user should disengage, remove the magnetic tip and load the wire sample. Specifically, the wire sample should be loaded with the horizontal wires parallel to the user. The new surface is focused upon and the tip engages for the second time. Common settings for the wire is: scan size = 20 to 50 pm; the lift height is 200 nm; the phase z-range is 2 to 5 degrees; the height z-range is 500 nm. For DC current imaging, the piezo is driven externally by a function generator and the current comes from a DC power supply. The piezo drive input is ANA 1. The lock-in references the sync output of the function generator and takes in the photodetector output from the SAM. For AC current imaging, the piezo is driven internally during the topographic scans and is not driven during the lift scans. The function generator provides the AC current to the wires and the reference for the lock-in. Data acquisition is enabled by Controller.vi. Appendix B Magnetic Field Over a Wire The cantilever response depends critically on the magnetic field produced by the straight conductor. This magnetic field is derived by first considering the case of a long wire with negligible width and height, which will be called a point wire. Secondly, the magnetic field of a wire with finite width and height, a bar wire, is presented. The law of Biot and Savart expresses the total magnetic field H at any point in space due to the current as H = - dx (B. 1) r2 47r where dl is a current element and r is the distance from the element to the observation point[13]. Applying it to the case of a straight conductor with current, I, and length, 2a (Fig. B-1), the magnetic field at point P can be solved, with r = H= I 4r I / 2dya(X2 'Y= afX + -f v/x 2 + y2 , dl = 47r (2+y)/ Eqn. B.2 is simplified if the conductor is very long, a H = dy, and sinG = X/ 2a x a xv'z2 + a2 X2 + y 2: (B.2) > x, -(B.3) 27rx Furthermore, as long as the observation point is at the same distance away from the wire, it expe109 x P r 0 -a dl a Figure B-1: Magnetic field produced by a straight conductor at point P. riences the same magnetic field. Hence Eqn. B.3 can be rewritten as H= (B.4) 27rr where r is the distance between the observation point and the wire. Because the MFM tip is usually magnetized in one specific direction, separating the x- and zcomponents of the magnetic field is necessary. As illustrated in Fig B-2, the direction of the magnetic field is related to the 0-component, H= = - (sin Oi - cos 0-) 27rr 27rr (B.5) It can also be re-expressed in cartesian-coordinates, H= 27r ( 2 (X2 + Z2) -2 (B.6) (X2 + Z2) The tip is most often magnetized in the vertical direction (z) because it reacts more sensitively to the vertical force. Therefore it is helpful to isolate the z-component of the magnetic field: Hz = -- 2 27r (x + z 2) Physcially, Hz is asymmetric about the z-axis and is maximized along the x-axis. (B.7) 111 Fig. B-3 illustrates a wire with finite width, W, and height, H. The total magnetic field can be expressed as the sum of the magnetic fields produced by each infinitissimal units over the crosssectional area, and the z-component is hence Hz = I I 2,rWH fW/2 r0 J-W/2 f-H xx (x - XO) 2 + (z dzdx (B.8) - ZO)2 Fig. B-4 shows the vertical components of H, VH, and V 2H as a function of lateral (x) and vertical (z) distance from a 2 pm-wide wire carrying 50 mA current. V 2H is directly related to the change in the cantilever's phase response as explained in Sec. 1.2. APPENDIX B. MAGNETIC FIELD OVER A WIRE 112 (x, z) Magnetic field,H - 0 / r X o 4 x Wire c arrying current, I, in +y direction Figure B-2: Magnetic field about a point wire with current, I, in the +y direction. z (x, z) (X zo) LulW Figure B-3: Magnetic field about a wire with width, W, and height, H. It is equivalent to the summation of the magnetic fields of infinitissimal pieces at (xo, zo) with width dx and height dz over the cross-sectional area. 113 500 a) 450 400 E 0 ,s350 N 300 50 250 -100 200 -2 500 -1.5 -1 0 -0.5 0.5 1 1.5 2 H in[Oe] b) 450 400 E 'S350 N 300 250 200 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 -1.5 -1 -0.5 0 x [um] 0.5 1 1.5 2 500 450 400 E , 350 N 300 250 200 -2 Figure B-4: The vertical components of a)H, b)VH, and c)V 2 H as a function of lateral (x) and vertical (z) distance from a 2 pm-wide wire carrying 50 mA current 114 APPENDIX B. MAGNETIC FIELD OVER A WIRE Appendix C Matlab Code C.1 DC Current Imaging C.1.1 DCcurrent.m % DCcurrent.mM % calculates magnitude and phase response to DC currentM % y represents cantilever deflectionM clear allM syms q yo M x-cant k M syms x z x zO IcM % transfer function U M 10 matrixA = [-cos(q)-cosh(q) -sin(q)-sinh(q); ... . M*(sin(q)-sinh(q))-cos(q)+cosh(q) M*(-cos(q)-cosh(q))-sin(q)+sinh(q)];M matrixB = [yo/2*(cos(q)-cosh(q)); yo/2*(cos(q)+cosh(q)-M*(sin(q)+sinh(q)))]; U = simple(matrixA -1*matrixB);M y = yo/2*(cos(k*x-cant)+cosh(k*x-cant)) U(1)*(cos(k*x-cant)-cosh(k*x-cant)) + ... M + U(2)*(sin(k*x-cant)-sinh(k*x.cant));I del-y = diff(y,x-cant,1);M M4 115 20 APPENDIX C. MATLAB CODE 116 E = 1.69e11; rho = 2330; a = 28.9e-6; b = 3.37e-6; L = 227.le-6; % Young's modulus [N/m^2]M x-cant % setting to y(x-cant=L) and dy(x-cant=L)/dx-cantM % amplitude of bimorph driveM yo = L; 10e-9; A = a*b; I = a*b^3/12; % density [kg/rn3]M % width of beam [m]M % thickness of cantilever [m]A % length of cantilever [m]k % cross section area of cantileverM % moment of inertiaM %################################################################## % solve for H fieldA M total current sent through wireM Ic = .001; width of wire sampleM W = 2e-6; height of wire sampleM H = 200e-9; magnetic dipole moment of tip in x M m-x = 0; m-z = 30 0; 40 magnetic dipole moment of tip in z [A*m^2] A! m-q = le-8; % magnetic monopole moment [Am]M u-0 = 4*pi*le-7; Q = 180; kc = 3*E*I/L-3 %permeability of free spaceM quality factorM effective spring constant of cantilever [N/m];M % function for z component of H field at one singular pointM func = Ic/(2*pi*W*H) * -(x-x0)/((x-x0)^2 + (z-z0)^2);M 50 % H field summed over width and heightA! Hz = int( int(func, x, -W/2, W/2), z, -H, 0);M % first derivative of Hz to find F-mag = m-z * dHz/dzM dHz = diff(Hz, z, 1); M ddHz = u-0*(m-z*dHz+m-q*Hz); M Fmag kmag diff(Hz, z, 2);M = diff(Fmag, z, 1); % kmag 60 = dFmag/dz evaluated at z-eqM points = 5001;M startx = -5*W;M endx = -startx;M x = [startx: (endx-startx)/points:endx];M C.1. DC CURRENT IMAGING heightabove = 800e-9; z = zeros(1,points);M 117 M 70 counter = 1;M for dummy = startx:(endx-startx)/points:-W/2M z(counter) = heightabove-H;M counter = counter + 1; M endM for dummy = -W/2:(endx-startx)/points:W/2K4 z(counter) = heightabove;M counter = counter + 1;M endM 80 for dummy = W/2: (endx-startx)/points:endxM z(counter) = heightabove-H;M counter = counter + 1;M endM z = z(1:length(x));M num-Hz = eval(vectorize(Hz));M num-kmag = eval(vectorize(kmag));M 90 % for Point Mass ModelM num-ddHz = eval(vectorize(ddHz));M phi = Q/kc*m-z.*num-ddHz;M %##############################################################A wO = (1.8751041)^2/L^2*srt(E*b^2/12/rho)*.5*(sqrt(4-1/Q^2));M % natural resonance of fundamental mode M fO = wO/2/pi M % [Hz]M 100 w = wO;m % driving at first resonant freq.M w-k = 2*w/(sqrt(4-1/Q^2)+j/Q);M damping = w-k/Q;M '(4 k = sqrt(w-k*sqrt(rho*A/E/I));4 M q 110 = k*L;M M = kc*q^3/3./num-kmag;M APPENDIX C. MATLAB CODE 118 num-y = eval(vectorize(y));M num-deLy = eval(vectorize(de_y)); mass = kc/wO^2;M PM = 4.9960e+003./ (w0^2-w^2+num_kmag/mass + j*w*wO/Q);M 120 % Plotting of FiguresM figureM superplot(x*1e6,180/pi*(angle(numy) -angle(num-y(1))))M ylabel('Phase [deg]')M xlabel('x [um]')M AC Current Imaging C.2 ACcurrent.m C.2.1 % ACcurrent.mM % calculates magnitude and phase response to AC currentM % z-ca represents cantilever deflectionM 4 clear all % transfer function UM K4 syms k-ca L dcF E I x-cantM 10 K4 -sin(kca*L)-sinh(kca*L); ... M matrixA = [-cos(kca*L)-cosh(kca*L) (sin(kca*L)-sinh(kca*L)) (-cos(k_ca*L)-cosh(k-ca*L))];M matrix-B = [0; dcF/E/I/k_ca^3];M U = (matrixA^-1*matrixB);M 4 z-ca = U(1)*(cos(k-ca*x-cant)-cosh(k-ca*x-cant)) + ... 4 U(2)*(sin(k-ca*x-cant)-sinh(k-ca*x-cant));M M 20 % cantilever parametersM E = 1.69e11; rho = 2330; % Young's modulus [N/m2]M % density [kg/m^3]M C.2. AC CURRENT IMAGING a b = L = 28.9e-6; 3.37e-6; 227.le-6; 119 % width of beam [m]M % thickness of cantilever [m]k % length of cantilever [m]M x-cant = L; K4 30 A = a*b; I = a*b^3/12; % cross section area of cantileverM % moment of inertiaM 4 % wire parameters and magnetic fieldM 4 syms x z x zA M Ic = .002; W = 2e-6; H 200e-9; m.x = 0; m-z = 0; q = 2.8e-5; % total current sent through wireM % width of wire sampleM % height of wire sampleM 40 % magnetic moment of tip in x M % magnetic moment of tip in z M % magnetic monopole momentM u-0 = pi*4e-7;M Q = 180; kc 3*E*I/L^3; % quality factorM % effective spring constant of cantilever [N/m];M M4 % function for z component of H field at one singular pointM func = Ic/(2*pi*W*H) * -(x-x0)/((x-x0)^2 + (z-z0)^2);M4 50 % H field summed over width and heightM Hz = int( int(func, x, -W/2, W/2), zO, -H,0);M % first derivative of Hz to find F-mag dHz = diff(Hz, z, 1); M = m-z * dHz/dzM 4 Fmag 4 kmag u_0*(m-z*dHz+q*Hz); M = diff(Fmag, z, 1);K4 60 K4 points = 2002;M startx -12.5*W;M endx = -startx;m K4 x = [startx: (endx-startx) /points:endx] ;4 4 heightabove = 8400e-9; z = zeros(1,points);- M APPENDIX C. MATLAB CODE 120 70 counter = 1;M for dummy = startx: (endx-startx)/points: -W/2M z(counter) = heightabove-H;M counter = counter + 1; M endM for dummy = -W/2:(endx-startx)/points:W/2M z(counter) = heightabove;M counter = counter + 1;K endM M 80 for dummy = W/2:(endx-startx)/points:endxM z(counter) = heightabove-H;M counter = counter + 1;K endM z = z(1:length(x));M numHz = eval(vectorize(Hz));M % case of AC currentM 90 dcF = eval(vectorize(Fmag));M %##################################################################$t % drive frequencyM t = 0;M wO = (1.8751041)^2/L^2*srt(E*b^2/12/rho)*.5*(sqrt(4-1/Q^2));M % natural resonance of fundamental mode M wc = wO;M 100 fc = wc/2/piM w-ca = 2*wc/(sqrt(4-1/Q^2)+j/Q);M k-ca = sqrt(w-ca*sqrt(rho*A/E/I));M num-z-ca = eval(vectorize(z-ca));K %##################################################################$ % plotsM M figureno = 2M if figureno == 1M figureM 110 C.2. AC CURRENT IMAGING 121 subplot(2,1,1)M splot(x*1e6, abs(num-z-ca)*1e9);M xlabel('x [um]')M ylabel('Magnitude [nm]');M subplot(2,1,2)M splot(x*1e6, angle(num-z-ca)*180/pi);M xlabel('x [um]')M ylabel('Phase 120 [deg]');M M elseif figureno == 2M newzca(1:points/2) = -abs(num-z-ca(:points/2))*e9-Q newzca(points/2+1:points+1) = abs(num-z ca(points/2+1:points+1))*1e9;M splot(x*1e6,newzca/1000, '--')M axis tightM endM C.2.2 decay.m % decay.mM % shows decay of Hz and dHzM % called by decay-use.mM function [numHz, num-dHz] = decay(num-z)M E = 1.69e11; rho = 2330; a = 28.9e-6; b = 3.37e-6; L = 227.le-6; A = a*b; I = a*b^3/12; I .002; W = 2e-6; H = 200e-9; Q = 180; kc = 3*E*I/L^3; % % % % Young's modulus [N/m^2]M density [kg/rn ^3]M width of beam [m]M thickness of cantilever [m]M 10 % length of cantilever [m][ % cross section area of cantileverM % moment of inertiaM % total current sent through wireM % width of wire sampleM % height of wire sampleM % quality factorM % effective spring constant of cantilever [N/m];M 20 x = sym('x');M z = sym('z');4 x = sym('xO');M zA = sym('zO');M 4 % function for z component of H field at one singular pointM APPENDIX C. MATLAB CODE 122 func = I/(2*pi*W*H) * -(x-xO)/((x-xO)^2 + (z-z0)^2);M % H field summed over width and heightM Hz = int( int(func, x, -W/2, W/2), zA, -H, 0);M 30 % second derivative of Hz to find F-mag = m-z*ddHzM ddHz = diff(Hz, 'z', 2);M dHz = diff(Hz, 'z', 1);K startx = -5*W;M endx = -startx;M points = 599;M 40 x = [startx:(endx-startx)/points:endx];M z = num_z;M numHz = eval(vectorize(Hz));M num-dHz = eval(vectorize(dHz));M num-ddHz = eval(vectorize(ddHz));M C.2.3 decayuse.m % decay.mM % shows decay of Hz and dHzM % called by decay-use.mM function [num-Hz, num-dHz] = decay(num-z)M E = 1.69e11; rho = 2330; a = 28.9e-6; b = 3.37e-6; L = 227.le-6; A = a*b; I = a*b^3/12; Young's modulus [N/m^2]M density [kg/m^3]1 width of beam [m]M thickness of cantilever [m]M length of cantilever [m]k I total current sent through wireM width of wire sampleM = .002; W = 2e-6; H = 200e-9; Q = 180; kc = 3*E*I/L^3; 10 cross section area of cantileverM moment of inertiaM height of wire sampleM quality factorM effective spring constant of cantilever [N/m];M 20 x = sym('x');M C.3. NON-LINEAR CURRENT IMAGING 123 z = sym('z');M x = sym('xO');M zO = sym('zO');M % function for z component of H field at one singular pointM func = I/(2*pi*W*H) * -(x-xO)/((x-xO)^2 + (z-zO)2);M % H field summed over width and heightM Hz = int( int(func, x, -W/2, W/2), zO, -H, 0);M I4 30 % second derivative of Hz to find F-mag = m-z*ddHzM ddHz = diff(Hz, dHz = diff(Hz, 'z', 2);M 'z', 1);M I4 startx = -5*W;M endx = -startx;m points = 599;K M 40 x = [startx:(endx-startx)/points:endx]; z = num-z;M numHz = eval(vectorize(Hz));M numdHz = eval(vectorize(dHz));M num-ddHz = eval(vectorize(ddHz));M use.tex C.3 Non-linear Current Imaging C.3.1 degenerate.m % degenerate.mM % calculates magnitude and phase response to degenerate non-linearM clear allM % cantilever parametersM K4 E = 1.69e11; % Young's modulus [N/m-2]M rho = 2330; % density [kg/m3]M a = 28.9e-6; % width of beam [m]M b = 3.37e-6; % thickness of cantilever [m]M L = 227.le-6; % length of cantilever [m]m 10 APPENDIX C. MATLAB CODE 124 x-cant = L; % setting to y(x-cant=L) and dy(x- cant=L)/dx-cantM yo % amplitude of bimorph driveM Oe-9; A = a*b; I a*b^3/12; M~ % cross section area of cantileverM % moment of inertiaM 20 % wire parameters and magnetic fieldM syms x z x zA M Ic = 0.002; W = 2e-6; H = 200e-9; % total current sent through wireM % width of wire sampleM mx % magnetic dipole moment of tip in x M % magnetic dipole moment of tip in z M = 0; m-z = 0; % height of wire sampleM % magnetic monopole moment M q = 8e-6; u-0 = pi*4e-7;M % quality factorM Q = 180; kc = 3*E*I/L-3; % effective spring constant of cantilever [N/m];M K4I % function for z component of H field at one singular pointM func = Ic/(2*pi*W*H) * -(x-x0)/((x-x0)^2 + (z-z0)^2);M % H field summed over width and heightM Hz = int( int(func, x, -W/2, W/2), zA, -H,0);M M % first derivative of Hz to find Fmag = m-z * dHz/dzM dHz = diff(Hz, z, 1); M 30 40 u0*(m-z*dHz+q*Hz); M Fmag = kmag = diff(Fmag, z, 1); % kmag = dFmag/dz evaluated at z-eqM points = 5001;K endx = 10*W;M startx = -endx;M 50 x = [startx: (endx-startx)/points:endx];M heightabove = 1000e-9; z = zeros(1,length(x));M counter = 1;M M C.3. NON-LINEAR CURRENT IMAGING 125 for dummy = startx:(endx-startx)/points:-W/2M z(counter) = heightabove-H;M counter = counter + 1; endM M 60 for dummy = -W/2:(endx-startx)/points:W/2M z(counter) = heightabove;M counter = counter + 1;K endM for dummy = W/2: (endx-startx)/points:endxM z(counter) = heightabove-H;M counter = counter + 1;M 70 endM z = z(1:length(x));M numHz = eval(vectorize(Hz));M A t ACcurrent = 1;M if ACcurrent ==IM % case of AC currentM sO = zeros(1,length(x));M 80 si = eval(vectorize(kmag));M dcF = eval(vectorize(Fmag));M elseM % case of DC currentM sO = eval(vectorize(kmag));M si = zeros(1,length(x));K dcF = zeros(1,length(x));M endM 90 phi = O;M MO Ml cMl sO/(E*I);M sl*exp(j*phi)/(2*E*I);M sl*exp(-j*phi)/(2*E*I);M %##################################################################A % drive frequencyM t = 0;M 100 %offset = 186.44*2*pi; % for magnitudeM offset % for phaseM = 0; APPENDIX C. MATLAB CODE 126 wO = (1.8751041)^2/L^2*sqrt(E*b^2/12/rho)*.5*(sqrt(4-1/Q^2));M % natural resonance of fundamental mode M w = wO-offset;M wpl = 2*w/(sqrt(4-1/Q^2)+j/Q);M w-nl = 2*w/(sqrt(4-1/Q^2)-j/Q);M w-p2 = 4*w/(sqrt(4-1/Q^2)+j/Q);M w-n2 = 4*w/(sqrt(4-1/Q^2)-j/Q);M k-pl k-nl k-p2 k-n2 110 = sqrt(w-pl*sqrt(rho*A/E/I));M = sqrt(w-nl*sqrt(rho*A/E/I));M = sqrt(w-p2*sqrt(rho*A/E/I));M = sqrt(w-n2*sqrt(rho*A/E/I));M y0 = zeros(1,length(x)); M y-pl = zeros(1,length(x));M y-nl = zeros(1,length(x));M y-p2 zeros(1,length(x));M y-n2 = zeros(1,length(x));M dr g zeros(1,length(x));M dely-p = delypl zeros(1,length(x));M delynl = zeros(1,length(x));M dely-p2 = zeros(1,length(x));M dely-n2 = zeros (1,lengt h(x));M for i = 1:length(x)M a3 a4 b2 b4 c2 c4 d2 d4 f2 f4A % 0 0;K 0 0 0 0 0 matrixA = [2 6*L 0 0 ... K -sin(k-nl*L)-sinh(k-nl*L) 0 -cos(k-nl*L)-cosh(k-n1*L) 0 0 0 0 0 0;K 0 -cos(k-p1*L)-cosh(k.p1*L) -sin(k.p1*L)-sinh(k-pl*L) ... M 0 0 0 0 0 0 0;K 0 ... M 0 0 0 0 0 -cos(k-n2*L) -cosh(k-n2*L) -sin(k-n2*L)-sinh(k-n2*L) 0 0;M 0 ... M 0 0 0 0 0 0 0 -cos(k-p2*L)-cosh(k-p2*L) -sin(k-p2*L)-sinh(k-p2*L);M M0(i)*L^2 M0(i)*L^3-6 ... M M1(i)*(cos(kLnl*L)-cosh(k-nl*L))...M M1(i)*(sin(k-nl*L)-sinh(k-nl*L)) ...A cMl(i)*(cos(k-pl*L)-cosh(k-pl*L)) ...M cMl(i)*(sin(kLpl*L)-sinh(k-pl*L)) ... 0 0 0 0;K 120 130 140 C.3. NON-LINEAR CURRENT IMAGING cM(i)/k-n1^3*L^2 cMl(i)/k-n1^3*L^3...M MO(i)/k-n1^3*(cos(k-nl*L)-cosh(k-nl*L))-sin(k-n1*L)+sinh(k-nl*L).. .M MO(i)/k-n1^3*(sin(k-nl*L)-sinh(k-nl*L))+cos(k-nl*L)+cosh(k-nl*L)...M o o ... M M1(i)/k-n1^3*(cos(k-n2*L)-cosh(k-n2*L))...M M1(i)/kn1^3*(sin(k-n2*L)-sinh(k-n2*L)).. .M o 0 0 127 150 0;m cMl(i)/k-n2^3*(cos(k-nl*L)-cosh(k-nl*L)) ... M cMl(i)/k-n2^3*(sin(k-nl*L)-sinh(k-n1*L)) ... M o 0 ... M MO(i)/k-n2^3*(cos(k-n2*L)--cosh(k-n2*L))-sin(k-n2*L)+sinh(k-n2*L)...M MO(i)/k-n2^3*(sin(k-n2*L) -sinh(k-n2*L))+cos(k-n2*L)+cosh(k-n2*L). .. 0 0;K M1(i)/k-pl^3*L^2 M1(i)/k-p1^3*L^3...D 0 160 ... m 0 MO(i)/k-p1^3*(cos(k-pl*L)-cosh(k-pl*L))-sin(k-pl*L)+sinh(k.pl*L).. .m MO(i)/k-p1^3*(sin(k-p1*L)-sinh(k-p1*L))+cos(k-p1*L)+cosh(k-p1*L)...M 0 0 0 0 ...m cMl(i)/k-p1^3*(cos(k-p2*L)-cosh(k-p2*L))...M cMl(i)/k-pl^3*(sin(k-p2*L)-sinh(k-p2*L));M 0 0 M1(i)/k-p2^3*(cos(k-p1*L)-cosh(k-p1*L)).. .M M1(i)/k-p2^3*(sin(k-pl*L)-sinh(k-pl*L)) 0 0 ... M MO(i)/k-p2^3*(cos(k-p2*L)-cosh(k-p2*L))-sin(k-p2*L)+sinh(k-p2*L)...M MO(i)/k-p2^3*(sin(k-p2*L)-sinh(k-p2*L))+cos(k-p2*L)+cosh(k-p2*L)M 170 ];m matrix-B = [O;M yo/4*(cos(k-nl*L)-cosh(k-n1*L));M yo/4*(cos(k-pl*L)-cosh(k-pl*L));M o;M 0; K (-Ml(i)*yo/4*(cos(k-nl*L)+cosh(k-nl*L)) - ... M cMl(i)*yo/4*(cos(k-p1*L)+cosh(k-pl*L)));M yo/4*(sin(knl*L)+sinh(k-nl*L) - .M.S MO(i)/k-nl^3*(cos(k-nl*L)+cosh(k-nl*L)))-...4 dcF(i)/2/E/I/k-n1^3*exp(-j*phi); M -cMi(i)/k-n2^3*yo/4*(cos(k-nl*L)+cosh(k-nl*L));M yo/4*(sin(k-pl*L)+sinh(k-p1*L) - ... n MO(i)/k-p1^3*(cos(k-pl*L)+cosh(k-pl*L)))-...M dcF(i)/2/E/I/k-nl^3*exp(j*phi);M -Ml(i)/k-p2^3*yo/4*(cos(k-pl*L)+cosh(k-pl*L))M ];m M U = (matrix-A-1*matrix-B);M M4 180 190 APPENDIX C. MATLAB CODE 128 y_0(i) U(1)*x-cant^2 + U(2)*xcant^3;M = y-pl(i) = (yo/4*(cos(k-pl*x-cant)+cosh(k-pl*x-cant)) + U(5)*(cos(k-pl*x-cant)-cosh(k-pl*x-cant)) + . .M U(6)*(sin(k-pl*x-cant)-sinh(k-pl*x-cant)));M y-nl(i) = (yo/4*(cos(k-nl*x-cant)+cosh(k-nl*x-cant)) + U(3)*(cos(k.nl*x-cant)-cosh(k-nl*x-cant)) . + .M.. U(4)*(sin(k-nl*x-cant)-sinh(k-nl*x-cant))) ;M yp2(i) = (U(9)*(cos(k-p2*x-cant)-cosh(k-p2*x-cant)) + ... M U(10)*(sin(k-p2*x-cant)--sinh(k-p2*x..cant)));M y-n2(i) = (U(7)*(cos(k-n2*x-cant)-cosh(k-n2*x-cant)) + .M U(8)*(sin(k-n2*x-cant) -sinh(kn2*xcant)));M 200 delyO(i) dely-pl(i) = U(1)*2*x-cant + U(2)*3*x cant^2;M k-pl*(yo/4*(-sin(k-pl*x-cant)+sinh(kpl*x-cant)) + .M U(5) *(-sin(k-p1*x.cant)-sinh(k-pl*x.cant)) + ... M dely.nl(i) = knl*(yo/4*(-sin(k-nl*x-cant)+sinh(k-nl*x-cant)) + ... 1 U(3) *(-sin(k-nl*x.cant)-sinh(k-nl*x-cant)) + .. .M U(6)*(cos(k-pl*x-cant)-cosh(k-p1*x-cant)));M 210 U(4)*(cos(k-nl*x-cant)-cosh(k-nl*x-cant)));M dely-p2(i) = k-p2*(U(9)*(-sin(k-p2*x-cant)-sinh(k-p2*x-cant)) + .M.S U(10)*(cos(k-p2*x-cant)-cosh(k-p2*x-cant)));M dely-n2(i) = k-n2*(U(7)*(-sin(k-n2*x-cant)-sinh(k-n2*x-cant)) + .M.S U(8) *(cos(k-n2*xcant)-cosh(k-n2*x-cant)));M endM %##################################################################$ % plotsM 220 if ACcurrent == 1M figureM subplot(2,1,1)M plot(x*1e6, 2*abs(yp1)*1e9);M xlabel('x [um]')M ylabel('Magnitude [nm]');M subplot(2,1,2)M plot(x*1e6, angle(y-pl)*180/pi);M xlabel('x [um]')M ylabel('Phase [deg] ');M figureM subplot(2,1,1)M plot(x*1e6, 2*abs(yp2)*1e9);M xlabel('x [um]')N ylabel('Magnitude [nm]');M subplot(2,1,2)M 230 C.3. NON-LINEAR CURRENT IMAGING 129 plot (x*1e6, angle(y _p2)*180/pi);M xlabel('x [umx]')M ylabel('Phase [deg] ');M 240 elseM figureM subplot(2,1,1)M plot(x*1e6, 2*abs(y-pl)*1e9);M xlabel('x [um]')M ylabel('Magnitude [nim] ');M subplot(2,1,2)K plot(x*1e6, 180/pi*(angle(y-pl)-angle(y-pl(1))));M xlabel('x [um]')M 250 ylabel('Phase [deg] ');M endM C.3.2 nondegenerate.m % nondegenerate.mM % calculates cantilever response for wp \= wcM clear allM M E = 1.69e11; rho = 2330; a = 28.9e-6; b = 3.37e-6; L = 227.le-6; A I 12e-9; = a*b; a*b^3/12; syms x z x % density [kg/m 3]M % width of beam [m]M % thickness of cantilever [m]?[ % length of cantilever [m]M 10 '( % amplitude of bimorph driveM x-cant = L; yo % Young's modulus [N/rm2]M cross section area of cantileverM % moment of inertiaM % zO M 20 Ic = .001; W = 2e-6; H = 200e-9; m-x = 0; % total current sent through wireM % width of wire sampleM % height of wire sampleM % magnetic moment of tip in x M APPENDIX C. MATLAB CODE 130 m-z = 0; q = le-5; u-0 = pi*4e-7;M Q 180; % magnetic moment of tip in z M kc = 3*E*I/L^3; % effective spring constant of cantilever [N/m];M % magnetic monopole moment(I % quality factorM M % function for z component of H field at one singular pointM func = Ic/(2*pi*W*H) * -(x-x0)/((x-x0)^2 + (z-z0)^2);M 30 % H field summed over width and heightM Hz = int( int(func, x, -W/2, W/2), z, -H,0);M % first derivative of Hz to find F-mag = m-z * dHz/dzM dHz = diff(Hz, z, 1); M 40 Fmag = u-0*(m-z*dHz+q*Hz); M kmag = diff(Fmag, z, 1); % kmag = dFmag/dz evaluated at z-eqM points = 2001;M startx = -12.5*W;M endx = -startx;M x = [startx: (endx-startx)/points:endx];M 50 heightabove = 10OOe-9; M z = zeros(1,points);M counter = 1;M for dummy = startx:(endx-startx)/points:-W/2M z(counter) = heightabove-H;M counter = counter + 1; M endM for dummy = -W/2:(endx-startx)/points:W/2M z(counter) = heightabove;M counter = counter + 1;M endM for dummy = W/2:(endx-startx)/points:endxM z(counter) = heightabove-H;M counter = counter + 1;M endM z = z(1:length(x));M 60 C.3. NON-LINEAR CURRENT IMAGING num-Hz = eval(vectorize(Hz));M 131 70 phi = 0; si = eval(vectorize(kmag));M dcF = eval(vectorize(Fmag));M M1 M1 cMl = sl*exp(j*phi)/(2*E*I);M sl*exp(-j*phi)/(2*E*I);M4 %##################################################################A$ % drive frequencyM t = 0;K offset = 186.44*2*pi; % for magnitudeM %offset = 0; % for phaseM 80 wO = (1.8751041)^2/L^2*sqrt(E*b^2/12/rho)*.5*(sqrt(4-1/Q^2));M % natural resonance of fundamental mode M fO = wO/2/pi M % [Hz]M 90 w = wO;M choice = 2;M if choice == 1M wp = 50000*2*pi;M wc = w-wp;M elseif choice == 2M 100 wc = 100*2*pi;M wp = w-wc;M elseM wp = 1;M wc = w;M endM fp = wp/2/pi; fc = wc/2/pi;M M w-ca w-cb w.cc w-cd 110 = 2*wc/(sqrt(4-1/Q^2)+j/Q);M = 2*wc/(sqrt(4-1/Q^2)-j/Q);M = 4*wc/(sqrt(4-1/Q^2)+j/Q);M = 4*wc/(sqrt(4-1/Q^2)-j/Q);M APPENDIX C. MATLAB CODE 132 w/Q w-pa = 2*wp/(sqrt(4-1/Q^2)-j/Q);M w-pb = 2*wp/(sqrt(4-1/Q2)-j/Q);M w-cpa = 2*(wc~wp)/(sqrt(4-1/Q-2)+j/Q);MI w-cpd = 2*(wc+wp)/(sqrt(4-1/Q^2)-j/Q);M 120 if wp > wcM w-cpb = 2*abs(wc-wp)/(sqrt(4-1/Q^2)-j/Q);M w-cpc = 2*abs(-wc+wp)/(sqrt(4-1/Q^2)+j/Q); M elseM w-cpb = 2*abs(wc-wp)/(sqrt(4-1/Q^2)+j/Q);M w-cpc = 2*abs(-wc+wp)/(sqrt(4-1/Q^2)-j/Q); M endM k-ca = sqrt(w-ca*sqrt(rho*A/E/I));M4 130 k-cb = sqrt(w-cb*sqrt(rho*A/E/I));M k-cc = sqrt(w-cc*sqrt(rho*A/E/I));M k-cd = sqrt(w-cd*sqrt(rho*A/E/I));M k-pa = sqrt(w-pa*sqrt(rho*A/E/I));M k-pb = sqrt(w-pb*sqrt(rho*A/E/I));M kcpa = sqrt(wcpa*sqrt(rho*A/E/I));M k-cpb = sqrt(w-cpb*sqrt(rho*A/E/I));M k-cpc = sqrt(w-cpc*sqrt(rho*A/E/I));M k-cpd = sqrt(w-cpd*sqrt(rho*A/E/I));M z-dc z-ca = zeros(1,length(x)); 140 M zeros(1,length(x));K z-cb = zeros(1,length(x));K z-cc = zeros(1,length(x));M z-cd zeros(1,length(x));M z-pa = zeros(1,length(x));M z-pb = zeros(1,length(x));M z cpa = zeros(1,length(x));M z-cpb = zeros(1,length(x));M z-cpc = zeros(1,length(x));M z-cpd = zeros(1,length(x));M for i = 1:length(x)M m 150 C.3. NON-LINEAR CURRENT IMAGING 133 M 160 matrix-A = [ see Eqn. 4.11 ];M matrix.B = [O;M O;M Q;M O;M 0;M yo/4*(cos(k pa*L)-cosh(kpa*L));M yo/4*(cos(k-pb*L)-cosh(k-pb*L));M 0;m 170 0;m 0;M 0;M 0;m -dcF(i)/2/E/I*exp(j*phi);M -dcF(i)/2/E/I*exp(-j*phi);M ; M ; M k -pa^3*yo/4*(sin(k-pa*L)+sinh(k -pa*L));M k-pb^3*yo/4*(sin(k-pb*L)+sinh(k-pb*L)) ;M -M1(i)*yo/4*(cos(kpa*L)+cosh(k-pa*L));M -Ml(i)*yo/4*(cos(k-pb*L)+cosh(k-pb*L));M -cMI(i)*yo/4*(cos(k-pa*L)+cosh(k.pa*L));M -cMi(i)*yo/4*(cos(k-pb*L)+cosh(k-pb*L))I 180 ];' U = (matrixA^-1*matrix-B);M z-dc(i) = U(1)*x-cant^2 + U(2)*x-cant^3;M + ... M z-cb(i) = U(5)*(cos(k-cb*x-cant)-cosh(k-cb*x-cant)) + U(6)*(sin(k-cb*x-cant) -sinh(k-cb*x-cant));M ... K z-cc(i) = + ... M + ... z-ca(i) = U(3)*(cos(k-ca*x-cant)-cosh(k-ca*x-cant)) 190 U(4)*(sin(k-ca*x-cant) -sinh(k-ca*x-cant));M U(7)*(cos(k-cc*x-cant)-cosh(k-cc*x-cant)) U(8)*(sin(k-cc*x-cant) -sinh(k -cc*x-cant));M z-cd(i) = U(9)*(cos(k-cd*x-cant)-cosh(k-cd*x-cant)) U(10)*(sin(k-cd*x-cant) -sinh(k z-pa(i) = cd*x-cant));M yo/4*(cos(k-pa*x-cant)+cosh(k-pa*x-cant)) U(11)*(cos(k-pa*x-cant)-cosh(k-pa*x-cant)) + + ... K ... M 200 U(12)*(sin(k-pa*x-cant) -sinh(k.pa*x-cant));M z-pb(i) = yo/4*(cos(k-pb*x-cant)+cosh(k-pb*x-cant)) U(13)*(cos(k-pb*x-cant)-cosh(k-pb*x-cant)) + U(14)*(sin(k-pb*x-cant)-sinh(k-pb*x-cant));M .. + ... n APPENDIX C. MATLAB CODE 134 z-cpa(i) = U(15)*(cos(k-cpa*x-cant)-cosh(k-cpa*x-cant)) U(16)*(sin(k-cpa*x-cant) -sinh(k-cpa*x-cant));M z-cpb(i) = U(17)*(cos(k-cpb*x-cant)-cosh(k-cpb*x-cant)) U(18)*(sin(k-cpb*x-cant) -sinh(k-cpb*x-cant));M z-cpc(i) = U(19)*(cos(k-cpc*x-cant)-cosh(k-cpc*x-cant)) U(20)*(sin(kLcpc*x-cant) -sinh(k-cpc*x-cant));M z-cpd(i) = U(21)*(cos(k-cpd*x-cant)-cosh(k-cpd*x-cant)) U(22)*(sin(k-cpd*x-cant) -sinh(k-cpd*x-cant));M + + + + .. ... ... ... M M 210 M endM %##################################################################$ % plotsM figureno = 1;M 220 if figureno == IM %figureM if choice == 1M subplot(312)K splot(x*1e6, 2*abs(z-cpa)*1e9);M xlabel('x [um]')M ylabel('Magnitude [nm] ');M axis([-15, 15, 0, max(2*abs(z-cpa)*1e9)])M m 230 elseif choice == 2M subplot(313)M splot(x*1e6, 2*abs(z-cpa)*1e9);M xlabel('x [um]')M ylabel('Magnitude [nm] ');M axis([-15, 15, 0, max(2*abs(z-cpa)*1e9)])M elseM subplot(311)M splot(x*1e6, 2*abs(z.ca)*1e9);M xlabel('x [um]')M ylabel('Magnitude [run] ');M axis([-15, 15, 0, max(2*abs(z-ca)*1e9)])) endM elseif figureno == 3M figureM subplot(2,3,1)M plot(x*1e6, abs(z.dc)*1e9);M 240 C.3. NON-LINEAR CURRENT IMAGING xlabel('x [uml')N ylabel('Magnitude [rnm]');M 135 250 subplot(2,3,4)M plot(x*1e6, angle(z-dc)*180/pi);M xlabel('x [um]')M ylabel('Phase [deg] ');K subplot(2,3,2)M plot(x*1e6, 2*abs(z-ca)*1e9);M xlabel('x [um]')M ylabel('Magnitude [run] ');M subplot(2,3,5)M plot(x*1e6, angle(z-ca)*180/pi);M xlabel('x [um]')N ylabel('Phase 260 [deg] '-);4 subplot(2,3,3)M plot(x*1e6, 2*abs(z-cc)*1e9);P xlabel('x [um]')N ylabel('Magnitude [un]');M subplot(2,3,6)K plot(x*1e6, angle(z-cc)*180/pi);M xlabel('x [um]')M 270 ylabel('Phase [deg] ');K elseif figureno == 4M I4 figureM subplot(4,4,1)K plot(x*1e6, abs(z-dc));M subplot (4,4,5)M plot(x*1e6, abs(z-ca));M 280 subplot (4,4,6)M plot(x*1e6, abs(z-cb));M subplot(4,4,7)K plot(x*1e6, abs(z-cc));KI subplot(4,4,8)K plot(x*1e6, abs(z-cd));M subplot(4,4,9)M plot(x*1e6, abs(z-pa));1 subplot(4,4,10)M plot(x*1e6, abs(z-pb));M 290 APPENDIX C. MATLAB CODE 136 subplot(4,4,13)K plot(x*1e6, abs(z-cpa));M subplot(4,4,14)M plot(x*1e6, abs(z-cpb));M subplot(4,4,15)K plot(x*1e6, abs(z-cpc));M 300 subplot(4,4,16)K plot(x*1e6, abs(z-cpd));M figureM subplot(4,4,1)K plot(x*1e6, angle(z-dc));M subplot(4,4,5)M plot(x*1e6, angle(z-ca))M subplot(4,4,6)K plot(x*1e6, angle(z-cb));M 310 subplot(4,4,7)M plot(x*1e6, angle(z-cc));M subplot(4,4,8)K plot(x*1e6, angle(z-cd));M subplot(4,4,9)K plot(x*1e6, angle(z-pa));M subplot(4,4,10)M 320 plot(x*1e6, angle(z-pb));M subplot(4,4,13)M plot(x*1e6, angle(z-cpa));M subplot(4,4,14)K plot(x*1e6, angle(z-cpb));M subplot(4,4,15)M plot(x*1e6, angle(z-cpc));subplot(4,4,16)M plot(x*1e6, angle(z-cpd));M e4 end 330 Bibliography [1] T. Alvarez, S. V. Kalinin, and D. A. Bonnell. Magnetic-field measurements of current-carrying devices by force-sensitive magnetic-force microscopy with potential correction. Appi. Phys. Lett., 78:1005-1007, February 2001. [2] K. L. Babcock, V. B. Elings, J. Shi, D. D. Awschalom, and M. Dugas. Field-dependence of microscopic probes in magnetic force microscopy. Appl. Phys. Lett., 69:705-707, July 1996. [3] L. Belliard, A. Thiaville, S. Lemerle, A. Lagrange, J. Ferre, and J. Miltat. Investigation of the domain contrast in magnetic force microscopy. J. Appl. Phys., 81:3849-3851, April 1997. [4] G. Binnig, C. F. Quate, and C. Gerber. Atomic force microscope. Phys. Rev. Lett., 56:930-933, March 1986. [5] H-J Butt and M. Jaschke. Calculation of thermal noise in atomic force microscopy. Nanotechnology, 6:1-7, 1995. [6] A. N. Campbell, Jr. E. I. Cole, B. A. Dodd, and R. E. Anderson. Magnetic force mi- croscopy/current contrast imaging: A new technique for internal current probing of ics. Microelectronic Engineering, 24:11-22, 1994. [7] A. M. Chang, H. D. Hallen, L. Harriott, H. F. Hess, H. L. Kao, J. Kwo, R. E. Miller, R. Wolfe, J. van der Ziel, and T. Y. chang. Scanning hall probe microscopy. Appl. Phys. Lett., 61:19741976, October 1992. [8] S. H. Crandall and N. C. Dahl. An Introduction to the Mechanics of Solids. McGraw-Hill, New York, New York, first edition, 1959. [9] C. W. de Silva. Vibration: Fundamentals and Practice. CRC Press, New York, New York, first edition, 2000. 137 BIBLIOGRAPHY 138 [10] T. Goddenhenrich, H. Lemke, U. Hartmann, and C. Heiden. Magnetic force microscopy of domain wall stray fields on single-crystal iron whiskers. Appl. Phys. Lett., 56:2578-2580, June 1990. [11] R. D. Gomez, A. 0. Pak, A. J. Anderson, E. R. Burke, A. J. Leyendecker, and I. D. Mayergoyz. Quantification of magnetic force microscopy images using combined electrostatic and magnetostatic imaging. J. Appl. Phys., 83:6226-6228, June 1998. [12] P. Grutter, H. J. Mamin, and D. Rugar. Scanning Tunneling Microscopy II, pages 151-207. Springer, Berlin Heidelberg, 1992. [13] Halliday. Physics, pages 10-119. Addison-Wesley, Reading, Massachusetts, second edition, 1973. [14] U. Hartmann. The point dipole approximation in magnetic force microscopy. Phys. Lett. A: Mater. Sci. Process, 137:475-478, June 1989. [15] R. Hillenbrand, M. Stark, and R. Guckenberger. Higher-harmonics generation in tapping-mode atomic-force microscopy: Insights into the tip-sample interaction. Appl. Phys. Lett., 76:34783480, June 2000. [16] H. J. Hug, B. Stiefel, P. J. A. van Schendel, A. Moser, R. Hofer, S. Martin, H.-J. Guntherodt, S. Porthun, L. Abelmann, J. C. Lodder, G. Bochi, and R. C. O'Handley. Quantitative magnetic force microscopy on perpendicularly magnetized samples. J. Appl. Phys, 83:5609-5620, June 1998. [17] J. R. Kirtley and Jr. J. P. Wikswo. Scanning squid microscopy. Annu. Rev. Mater. Sci., 29:117-148, 1999. [18] L. Kong and S. Y. Chou. Quantification of magnetic force microscopy using a micronscale current ring. Appl. Phys. Lett., 70:2043-2045, April 1997. [19] J. Lohau, S. Kirsch, A. Carl, G. Dumpich, and E. F. Wassermann. Quantitative determination of effective dipole and monopole moments of magnetic force microscopy tips. J. Appl. Phys., 86:3410-3417, September 1999. [20] H. J. Mamin, D. Rugar, J. E. Stern, B. D. Teris, and S. E. Lambert. Force microscopy of magnetization patterns in longitudinal recording media. Appl. Phys. Lett, 53:1563-1565, October 1988. BIBLIOGRAPHY 139 [21] Y. Martin and H. K. Wickramasinghe. Magnetic imaging by "force microscopy" with 1000 aa resolution. Appl. Phys. Lett, 50:1455-1457, May 1987. [22] S. McVitie, R. P. Ferrier, J. Scott, G. S. White, and A. Gallagher. Quantitative field measurements from magnetic force microscope tips and comparison with point and extended charge models. J. Appl. Phys., 89:3656-3661, April 2001. [23] G. Meyer and N. M. Amer. Novel optical approach to atomic force microscopy. Appl. Phys. Lett., 53:1045-1047, September 1988. [24] G. Meyer and N. M. Amer. Simultaneous measurement of lateral and normal forces with an optical-beam-deflection atomic force microscope. Appl. Phys. Lett., 57:2089-2091, November 1990. [25] A. Moser, H. J. Hug, I. Parashikov, B. Stiefel, 0. Fritz, H. Thomas, A. Baratoff, H-J Guntherodt, and P. Chaudhari. Observation of single vortices condensed into a vortex-glass phase by magnetic force microscopy. Phys. Rev. Lett., 74:1847-1850, March 1995. [26] A. Oral, S. J. Bending, and M. Henini. Scanning hall probe microscopy of superconductors and magnetic materials. J. Vac. Sci. Technol. B, 14:1202-1205, March 1996. [27] R. B. Proksch, T. E.Schaffer, B. M. Moskowitz, E. D. Dahlberg, D. A. Bazylinski, and R. B. Frankel. Magnetic force microscopy of the submicron magnetic assembly in a magnetotactic bacterium. Appl. Phys. Lett., 66:2582-2584, May 1995. [28] C. A. J. Putman, B. G. De Grooth, N. F. Van Hulst, and J. Greve. A detailed analysis of the optical beam deflection technique for use in atomic force microscopy. J. Appl. Phys., 72:6-12, July 1992. [29] U. Rabe and W. Arnold. Atomic force microscopy at mhz frequencies. Ann. Phys., 3:589-598, 1994. [30] U. Rabe, K. Janser, and W. Arnold. Vibrations of free and surface-coupled atomic force microscope cantilevers: Theory and experiment. Rev. Sci. Instrum., 67:3281-3293, September 1996. [31] D. Rugar, H. J. Mamin, and P. Guethner. Improved fiber-optic interferometer for atomic force microscopy. Appl. Phys. Lett., 55:2588-2590, December 1989. BIBLIOGRAPHY 140 [32] D. Rugar, H. J. Mamin, P. Guethner, S. E. Lambert, J. E. stern, I McFadyen, and T. Yogi. Magnetic force microscopy: general principles and application to longitudinal recording media. J. Appl. Phys., 68:1169-1183, August 1990. [33] M. V. Salapaka, H. S. Bergh, J. Lai, A. Majumdar, and E. McFarland. Multi-mode noise analysis of cantilevers for scanning probe microscopy. J. Appl. Phys., 81:2480-2487, March 1997. [34] D. Sarid. Scanning Force Microscopy. Oxford University Press, New York, New York, second edition, 1994. [35] S. D. Senturia. Microsystem Design. Kluwer Academinc Publishers, Boston, Mass., first edition, 2001. [36] R. W. Stark, T. Drobek, and W. M. Heckl. Tapping-mode atomic force microscopy and phaseimaging in higher eigenmodes. Appl. Phys. Lett., 74:3296-3298, May 1999. [37] R. W. Stark and W. M. Heckl. Fourier transoformed atomic force microscopy: tapping mode atomic force microscopy beyond the hookian approximation. Surf. Sci., 457:219-228, 2000. [38] P. J. van Schendel, H. J. Hug, B. Stiefel, S. Martin, and H.-J. Guntherodt. A method for the calibration of magnetic force microscopy tips. J. Appl. Phys, 88:435-445, July 2000. [39] L. N. Vu and D. J. Van Harlingen. Design and implementation of a scanning squid microscope. IEEE Trans. Appl. Superd., 3:1918-1921, March 1993. [40] A. Yariv. Introduction to optical electronics. Holt, Rinehard and Winston, New York, New York, second edition, 1976. [41] Q. Zhong, D. Inniss, K Kjoller, and V. B. Elings. Fractured polymer/silica fiber surface studied by tapping mode atomic force microscopy. Surf. Sci., 290:L688-L692, June 1993. S