K.A. Bell

Surface and interface effects on ellipsometric spectra of crystalline Si

We present the first systematic investigation of the differences among reference-quality ellipsometrically measured pseudodielectric function 〈e〉 spectra of crystalline Si, which are nominally used to approximate the bulk dielectric function of this material. In addition to the expected influence of residual overlayers, we identify surface-local-field and energy-derivative effects, the latter representing shifts between bulk and measured critical point energies, as well as changes in excited-carrier lifetimes due to the surface. Model calculations indicate that these four effects account for nearly all differences among spectra studied, although a second-energy-derivative component appears at the E1 transition in some cases. The isotropic contribution to the surface-local-field effect is observed for the first time.

Evidence of near-surface localization of excited electronic states in crystalline Si

Surface- and interface-related spectra, obtained either directly by techniques such as reflectance-difference (-anisotropy) spectroscopy or indirectly by subtracting pseudodielectric function spectra obtained ellipsometrically on surfaces with different chemical termination, exhibit features related to energy derivatives of the bulk dielectric function. We argue that these spectra provide direct evidence that the excitations involved are localized both in space and time. These data unequivocally indicate that critical point energies obtained from above-band-gap ellipsometric or reflectrometric optical spectra are not necessarily equal to bulk values, and that surface chemical and structural termination is at least one contributing factor. Present surface-optical calculations do not include these effects, which may explain, in part, remaining discrepancies between theory and experiment.

In situ monitoring of MOVPE growth by combined spectroscopic ellipsometry and reflectance-difference spectroscopy

Comprehensive characterization of epitaxial growth by metal organic vapor phase epitaxy (MOVPE) requires a combination of thin-film, near-surface-, and surface-sensitive techniques to determine layer thicknesses and compositions, composition of the most recently deposited material, and surface chemistry, respectively. These data can be obtained non-destructively by spectroscopic ellipsometry (SE) and reflectance-difference (-anisotropy) spectroscopy (RDS/RAS). Here we describe the first unified optical system, basically a rotating-polarizer ellipsometer (RPE) integrated into a modified commercial rotating-sample MOVPE reactor, that performs both SE and RDS simultaneously with a single optical path. Data are obtained in parallel from 240 to 840 nm with a high-speed 16-bit photodiode array (PDA) at a repetition rate greater than 2 Hz and a precision of ±0.0001. We provide examples of its use, and show in particular that GaP intermixes with Si during the initial stages of heteroepitaxy. Capabilities of the presented configuration and its potential for future investigations are discussed.

Real-time optical characterization of heteroepitaxy by organometallic chemical vapor deposition

Heteroepitaxy of GaP on Si(100) and GaAs(100) is investigated under organometallic chemical vapor deposition conditions using combined spectroscopic ellipsometry (SE) and non-normal-incidence reflectance-difference (-anisotropy) spectroscopy. Real-time monitoring greatly assists in identifying optimum starting surfaces for heteroepitaxy since prolonged exposure to PH3 results in roughening of Si(100) and GaAs(100) surfaces, in agreement with previous work. Real-time SE data of GaP on Si indicate that under our conditions GaP and Si interpenetrate as optically identifiable materials over the first 75 A, suggesting that either trimethylgallium or a reaction by-product can act as a catalyst for the formation of Si{111} facets.

Systematic differences among nominal reference dielectric function spectra for crystalline Si as determined by spectroscopic ellipsometry

Abstract Results of the first systematic investigation of differences among reference-quality ellipsometrically measured pseudodielectric function 〈e〉 spectra of crystalline Si are discussed. These data are nominally used to approximate the bulk dielectric function of this material for optical modeling. In addition to the expected influence of residual overlayers, we identify surface-local-field and energy-derivative effects even between spectra obtained for seemingly identical H-terminated surfaces. Model calculations indicate that these effects account for nearly all differences among spectra studied. The isotropic contribution to the surface-local-field effect is also reported.

Photon-induced localization and final-state correlation effects in optically absorbing materials

Two consequences of the absorption of light in optically absorbing materials that appear not to have been recognized previously are: (1) localization of the final electron and hole states involved in the absorption process into wave packets and (2) propagation of these wave packets with their respective group velocities. We demonstrate the existence of these phenomena by applying first-order time-dependent perturbation theory to a simple model that can be solved analytically even when correlations that are ordinarily discarded in the random phase approximation are retained. This approach provides a natural explanation of components in surface- and interface-optical spectra that are related to energy derivatives of the bulk dielectric function eb and apparent differences in nominally bulk critical point energies Eg and broadening parameters Γ depending on surface conditions.

PHOTON-INDUCED LOCALIZATION IN OPTICALLY ABSORBING MATERIALS

Abstract We show that components of surface- and interface-related optical spectra that are related to derivatives of their bulk dielectric functions are due to a dynamic photon-induced localization of the initial and final states. Localization is described by correlation effects that arise from the finite penetration depth of light in optically absorbing materials, and lead to a substantially different perspective of optical absorption than that given by conventional theory.

Investigation and Control of MOVPE Growth by Combined Spectroscopic Ellipsometry and Reflectance‐Difference Spectroscopy

The detailed characterization of epitaxial growth by metal organic vapor phase epitaxy (MOVPE) and its closed-loop feedback control at the sample level require a combination of thin-film, near-surface, and surface-sensitive techniques to determine layer thicknesses and compositions, the composition of the most recently deposited material, and surface chemistry, respectively. These data can be obtained nondestructively by spectroscopic ellipsometry (SE) and reflectance-difference (-anisotropy) spectroscopy (RDS/RAS). We describe the first unified optical system, basically a rotating-polarizer ellipsometer (RPE) integrated within a modified commercial rotating-sample MOVPE reactor, that performs both SE and RDS simultaneously in a single optical path. We provide examples of its use, showing in particular that GaP can intermix with Si during the initial stages of heteroepitaxy, and demonstrating sample-driven closed-loop feedback control of epitaxy through the fully automatic deposition of an InxGa1—xP parabolic quantum well. These results illustrate capabilities of the presented configuration and its potential for future use.

Interpretation of critical point energy shifts in crystalline Si by near-surface localization of excited electronic states

Abstract Various surface-optical spectra can be described in terms of energy derivatives of the bulk dielectric function. The spectra unequivocally indicate that critical point energies obtained from optical data are not necessarily equal to bulk values and that surface chemical and structural terminations are at least contributing factors. We invoke localization and transition-lifetime arguments to describe these effects. Existing surface-optical calculations do not address these contributions, which may explain in part why discrepancies remain between theory and experiment.