A Detailed Examination of Time-Resolved Pulsed Raman Temeperature Measurements of Laser Annealed Silicon
- PDF / 371,315 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 75 Downloads / 190 Views
A DETAILED EXAMINATION OF TIME-RESOLVED PULSED RAMAN TEMEPERATURE MEASUREMENTS OF LASER ANNEALED SILICON G. E. JELLISON, JR., D. H. LOWNDES, AND R. F. WOOD Solid State Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830 ABSTRACT Raman temperature measurements during pulsed laser annealing of Si by Compaan and co-workers are critically examined. It has been shown previously that the Stokes to anti-Stokes ratio depends critically upon the optical properties of silicon as a function of temperature. These dependences, coupled with the large spatial and temporal temperature gradients normally found immediately after the high reflectivity phase, result in large variations in the calculated temperature depending upon the probe laser pulse width and the pulse-to-pulse and spatial variations in the annealing pulse energy density. INTRODUCTION Pulsed laser annealing is proving to be a useful technique for many semiconductor device processing applications. From a more fundamental standpoint, the nonequilibrium thermodynamic processes which occur during the laser annealing process are also currently of great interest [1]. Two competing models have been presented to explain the phenomena associated with pulsed laser annealing: 1) the melting model [2], wherein the laser annealing process is described in classical thermodynamic terms, and the plasma annealing model [3], wherein the annealing behavior is believed to take place via the electron-hole plasma created by the laser beam. The melting model successfully describes detailed results of many experiments which imply that the surface region of the sample melts; these include time-resolved measurements of optical reflectivity and transmission, electrical conductivity and x-ray diffraction studies as well as post-annealing dopant profile measurements [1,4]. Compaan and co-workers [5], however, have published results of pulsed Raman experiments which are in apparent conflict with the predictions of the thermal melting model. In a Raman scattering experiment, one can measure the light emitted after either the emission (Stokes process) or the absorption (anti-Stokes process) of a phonon. The phonon population, determined by Bose-Einstein statistics, is given by no(wo,T) = [exp(Awo/kT-1I, where wo is the phonon frequency (Awo = 0.064 eV for the F25 , optical phonon in Si) and T is the lattice temperature. Therefore, the ratio of the intensities of the Stokes and anti-Stokes components (R(S/AS)) can be extremely T-dependent. The determination of T from R(S/AS), however, is not a simple matter. Wood and co-workers [6] have pointed out that 1) the T-dependence of the optical properties of Si, 2) the large T gradients that are present during laser annealing, and 3) the statistical nature of the experiments of Ref. 5 make a simplified treatment of the experimental Raman data difficult to justify. In a more recent work, Jellison et al. [7] have shown in detail the effects of the T-dependent optical properties of silicon on R(S/AS). This paper presents a *Research sponsored
Data Loading...
