Chemical and Structural Characterization of Ultrathin Dielectric Films using AEM

  • PDF / 2,798,419 Bytes
  • 6 Pages / 415.8 x 637.2 pts Page_size
  • 44 Downloads / 256 Views

DOWNLOAD

REPORT


J.H.J. SCOTT, E.S. WINDSOR

t Surface and Microanalysis Science Division, National Institute of Standards and Technology (NIST), Gaithersburg, MD 20899-8371, [email protected]

ABSTRACT The structure of ultrathin silicon oxynitride films, used as gate dielectrics in integrated circuits (ICs), is studied using analytical electron microscopy (AEM). Laterally homogeneous blanket films approximately 2 nm in thickness are characterized in cross section using a 300 keV field emission TEM/STEM. High resolution imaging (HRTEM) is used to investigate the accuracy and precision of film thickness measurements and their comparability to other techniques such as secondary ion mass spectrometry, spectroscopic ellipsometry, x-ray reflectivity, x-ray photoelectron spectroscopy, and medium energy ion scattering. A two dimensional magnification calibration scheme that fits a pair of basis vectors to experimental images is presented, and integrated intensity profiles are used to define film boundaries for measurement. These image processing tools simultaneously improve the repeatability of the measurements and remove subjective operator bias from the measurement process. INTRODUCTION The smallest feature that needs to be accurately characterized in ICs is the thickness of the gate dielectric. The most recent National Technology Roadmap for Semiconductors (NTRS) predicts gate oxide equivalent thicknesses below 2 nm in 2006 (in the 100 nm technology generation), and dielectric thicknesses less than I nm before 2012 [1]. To support the statistical process control (SPC) methodologies used in production wafer fabrication, the thickness must be

measured exactly: the precision of the measurement must be 10 times smaller than the acceptable tolerance for the fabrication process [2]. At the 6; level the gate oxide tolerance is 8% of the film thickness, so a precision-to-tolerance ratio of 0.1 requires that a 2 nm thick oxide be measured with a precision of approximately 0.02 nm [3]. Very few techniques enjoy this level of precision, especially when required to measure many samples per hour during in-line process monitoring. The current trend is to use a rapid and precise indirect metrology tool (such as ellipsometry) that has been calibrated against an array of other techniques, including TEM, secondary ion mass spectrometry (SIMS), medium energy ion scattering (MEIS), x-ray photoelectron spectroscopy (XPS), x-ray reflectivity (XRR), and electrical measurements such as capacitance-voltage (C-V) curves. Unfortunately, these techniques often disagree about a given film's thickness by amounts larger than their individual precisions. Since the different techniques are measuring different physical phenomena, the spread in their measurements is not surprising. Indeed, some argue that the question is ill-posed since we do not have an adequate definition of film thickness at the 0.01 nm length scale. t Certain commercial equipment, instruments, or materials are identified in this paper in order to specify the experimental procedure adequately