Extra steps have been found with atomic-force microscopy generated on rather flat silicon ͑100͒ surfaces annealed at 1200°C in hydrogen, through a comparison with the surface annealed in argon, which exhibits a typical ͑S a ϩS b ͒ step structure. It is suggested that the extra steps are spontaneously generated due to the relaxation of the strain energy associated with the atomic dimers on the reconstructed surface.

A vicinal silicon (111) surface exhibits well defined single steps after being annealed at 1200 °C in hydrogen, which is in sharp contrast with step bunches featuring the surface annealed in argon. As a temporary explanation for its ability to unzip the step bunches, we suggest that hydrogen destroys the faulted triangles of a [112̄] step, eliminates this kind of step, and eventually leaves the single [1̄1̄2] steps alone behind on the surface.

In this letter, we show that the misorientation of a vicinal (001) surface annealed in hydrogen is accommodated by a pair of near-Sa steps which are evolved in two perpendicular {110} independently and coordinately. We demonstrate that the deviation of surface orientation in a tiny area can be readily evaluated by inspecting the local variation of the near-Sa step pair. We finally indicate that the atomic steps on a surface annealed in argon exhibit in contrast, no conformity with the misorientation towards an azimuth other than {110}.

Nondestructive short wavelength laser scattering tomography ͑LST͒ is proposed for determining the depths and dimensions of microdefects to a depth of several micrometers in the subsurface region of silicon wafers. Measurements were made at two different temperatures, i.e., 23 and 83°C, utilizing the temperature dependence of the absorption coefficient of silicon at the laser wavelength ͑680 nm͒. From a comparison between the depths of the defects determined by this new LST method and those by transmission electron microscopy, it is verified that the depths of microdefects less than 5 m deep are determined well by our method.The quality of the subsurface region ͑to a depth of a few micrometers͒ in silicon wafers is one of the dominant factors for successful ULSI fabrication. The dimensions and densities of microdefects in this region are very important to the quality. Nondestructive methods have, therefore, been required for detecting subsurface microdefects. Short wavelength laser scattering tomography ͑S-LST͒ was developed to detect these defects, but it is incapable of measuring the depths and dimensions of individual defects. 1 An optical shallow defect analyzer is very useful for detecting microdefects in the subsurface region to a depth of 0.5 m, but the measurable region is too shallow. 2 In this paper, we propose an original method using nondestructive laser scattering tomography for measuring the depths and dimensions of microdefects in the region to a depth of several micrometers, utilizing the temperature dependence of the absorption coefficient of silicon at the laser wavelength. We demonstrate the depth determination of subsurface defects.When a laser beam with intensity, I 0 , and energy larger than the silicon bandgap ͑1.12 eV at 300 K͒ 3 is incident on the surface of a silicon wafer close to the Brewster angle ͑75.3°͒, 1,4,5 the intensity, I, of the scattered light due to a microdefect at depth d beneath the surface is given bywhere is the differential scattering cross section, t i is the energy transmittance of the incident laser through the air/silicon interface close to the Brewster incidence, t s is the energy transmittance of the scattered light through the silicon/air interface at normal incidence, h is the instrument function, and ␣(T) is the absorption coefficient ͑at the laser wavelength͒ as a function of the sample temperature, T. Note that the scattered light is observed just above the sample wafer.Although the optical path of the incident laser from the incident surface to the defect is approximately 4% longer than the depth, d, of the defect due to the incidence close to the Brewster angle, we neglect the difference in deriving Eq. 4, because the difference is small. If the scattering intensity, I, is measured at two different temperatures, T 1 and T 2 , the measured intensities, I 1 and I 2 , are expessed as I 1 ϭ ͑ I 0 t i t s h ͒ exp͑Ϫ2␣͑T 1 ͒d ͒ ͓2͔ I 2 ϭ ͑ I 0 t i t s h ͒ exp͑Ϫ2␣͑T 2 ͒d ͒ ͓3͔The d and the quantity A (ϭI 0 t i t s h) are derived as follows d ϭ ln͑I 1 /I 2 ͒/2͓␣͑ T ...

The outdiffusion of boron, antimony, and phosphorus from the bare silicon wafer at 1200 °C, especially its dependence upon the annealing atmosphere, has been studied with spreading resistance and secondary ion mass spectroscopy (SIMS). It is found that the boron outdiffusion proceeds when the crystal is annealed in hydrogen, but is completely suppressed in argon even if the doping concentration is as high as 3×1018 cm−3 and the annealing time is as long as 2 h. The dramatic dependence upon the atmosphere has not been observed for the other impurities and is temporarily related with the desorption process of boron atoms from the surface.