Towards Automated Nanoassembly With the Atomic Force Microscope: A Versatile Drift Compensation Procedure

[+] Author and Article Information
Florian Krohs1

Division Microrobotics and Control Engineering, University of Oldenburg, Oldenburg 26129, Germanyflorian.krohs@uni-oldenburg.de

Cagdas Onal

Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213cagdas@cmu.edu

Metin Sitti

Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213msitti@andrew.cmu.edu

Sergej Fatikow

Division Microrobotics and Control Engineering, University of Oldenburg, Oldenburg 26129, Germanyfatikow@uni-oldenburg.de

type: N1B, manufacturer: NIIFP, Moscow


Corresponding author.

J. Dyn. Sys., Meas., Control 131(6), 061106 (Nov 06, 2009) (8 pages) doi:10.1115/1.4000139 History: Received June 14, 2008; Revised May 07, 2009; Published November 06, 2009; Online November 06, 2009

While the atomic force microscope (AFM) was mainly developed to image the topography of a sample, it has been discovered as a powerful tool also for nanomanipulation applications within the last decade. A variety of different manipulation types exists, ranging from dip-pen and mechanical lithography to assembly of nano-objects such as carbon nanotubes (CNTs), deoxyribonucleic acid (DNA) strains, or nanospheres. The latter, the assembly of nano-objects, is a very promising technique for prototyping nanoelectronical devices that are composed of DNA-based nanowires, CNTs, etc. But, pushing nano-objects in the order of a few nanometers nowadays remains a very challenging, labor-intensive task that requires frequent human intervention. To increase throughput of AFM-based nanomanipulation, automation can be considered as a long-term goal. However, automation is impeded by spatial uncertainties existing in every AFM system. This article focuses on thermal drift, which is a crucial error source for automating AFM-based nanoassembly, since it implies a varying, spatial displacement between AFM probe and sample. A novel, versatile drift estimation method based on Monte Carlo localization is presented and experimental results obtained on different AFM systems illustrate that the developed algorithm is able to estimate thermal drift inside an AFM reliably even with highly unstructured samples and inside inhomogeneous environments.

Copyright © 2009 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

Topography image recorded in noncontact mode (scan area: 1.0×1.0 μm2) showing 15 nm sized Au nanoparticles on mica (Z axis is scaled up). The nanoparticles were manipulated with the AFM tip (pushing in contact mode) to form the letter “F.” When dealing with objects of this order of magnitude or smaller, spatial displacements due to thermal drift can affect the manipulation and have therefore be compensated for.

Grahic Jump Location
Figure 2

Height profiles utilized as sensor data for the drift tracking algorithm. Profiles are recorded at the same position immediately after each other with tip velocities of 10 μm/s. Because drift can be neglected in this small time frame, the deviation observable in the depicted line scans indicate the low signal-to-noise ratio of this kind of measurement.

Grahic Jump Location
Figure 3

Histogram displaying the absolute frequencies of the sum of squared differences calculated from height profiles successively recorded at the same position. 10,000 height profiles of 1 μm in length were measured, each with a velocity of 10 μm/s. The sum of squared differences was obtained by comparing each two consecutive profiles.

Grahic Jump Location
Figure 4

Example of a probability density function used as the motion model after convergence of the particle filter. Based on recent drift measurements, the expected drift vector vd since the last update of the particle is calculated. To incorporate uncertainties in drift velocity and direction, a Gaussian distribution is added to this drift vector.

Grahic Jump Location
Figure 5

Flowchart of the drift compensation algorithm

Grahic Jump Location
Figure 6

Topography images (10×10 μm2) of the Au/Si surface before (top left and bottom left (3D view and Z scaled)) and after (top right) the drift estimation (17 h later). To compare the results of the drift estimation algorithm, the upper topography images were cross-correlated and the maximum of this correlation was determined to be at X=247 nm and Y=−220 nm, which is mostly identical (ΔX=6 nm, ΔY=6 nm) to the results obtained by the particle filter algorithm.

Grahic Jump Location
Figure 7

Measured X and Y component of drift and temperature inside the AFM over a period of 17 h with the gold coated silicon substrate used as a sample

Grahic Jump Location
Figure 8

Initial 10×10 μm2 AFM scan where the user chooses separate regions for drift estimation (large square) and particle manipulation (small square)

Grahic Jump Location
Figure 9

Consecutive line scans on the same position without drift compensation

Grahic Jump Location
Figure 10

Consecutive line scans on the same position with drift compensation. At time t=30 min the estimated drift amounts to X=−112 nm and Y=237 nm. At time t=60 min drift is estimated to X=−168 nm and Y=190 nm.

Grahic Jump Location
Figure 11

AFM images of the same 5×5 μm2 area (a) before the experiment, (b) after the experiment, without drift compensation, and (c) after the experiment, with drift compensation

Grahic Jump Location
Figure 12

Measured drift over 60 min (on nanoparticle substrate)



Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In