Basics

Infrared (IR) spectroscopy forms part of the group of vibrational spectroscopy techniques, i.e. different vibrational modes of two or more bonded atoms (molecules) are detected. The mechanical analogon is a system of various particles bonded together by springs with respective force constants. The frequencies of the molecular vibrations happen to lie usually in the infrared region contrary to molecular rotational modes that may usually be found in the microwave range. Characteristic vibrational modes (ener-gies) then allow for the identification of different types of bonds and functional groups present in polyatomic molecules.

The existing selection rule for IR spectroscopy is that the electric dipole moment of the molecule has to change during vibration so that a particular vibration mode is infrared active. This implies that neither symmetric vibration modes can be detected by IR spectroscopy (no change in the dipole moment), nor homonuclear bonds as the latter do not exhibit electrical dipole moments at all. It should be mentioned that Ra-man spectroscopy, also belonging to the category of vibrational spectroscopy, is com-plementary to IR spectroscopy since the selection rule here is the change of the

electri-26 Characterization fundamentals

cal polarisability of the molecule, therefore being sensitive to symmetric and insensi-tive with respect to asymmetric vibration modes.

Fourier spectroscopy allows for the rapid scanning of the whole spectral range since no dispersive elements such as prisms or diffractive gratings are used. Here, the intensity – wavenumber correlation is determined by the computed Fourier transform of an experimentally extracted interferogram. The latter is generated by a setup based on the Michelson interferometer consisting of a beam splitter, a fixed mirror and a moving mirror. By moving the mirror and thereby creating a defined difference in the optical path length, the spectrum of the continuous light source is transferred into its interference spectrum. The beam then passes through the sample where selective ab-sorption takes place and the resulting intensity distribution (intensity vs. mirror posi-tion) is then detected. The theory of the Fourier spectroscopy implies that the exact spectrum may only be extracted from its interferogram if the latter was generated by an infinite traverse path x of the mirror and by reducing the measurement intervals Δx to infinitesimal small steps. In the experiment, the ideal conditions are furthermore dis-torted by wavelength dependent influences of the optical elements and the finite dy-namic of electrical conversion. The most important phenomena and consequences are summarized in [76].

The spectrum of a sample is generated by initially measuring the signal of the ap-paratus that is merely influenced by the radiation characteristics of the continuous light source and the entity of all optical elements. Then the sample of interest is positioned into the light path and the resulting signal is measured under essentially the same conditions. The division of both spectra eliminates the particular characteristics of the light source and the optics and hence results in the transmission spectrum of the sam-ple. Therefore the transmittance T is defined as

I0

T=I ,

(3-1) with I the intensity of light transmitted through the sample and I0 the intensity of light incident on the sample. Note that, in IR spectroscopy, the term wavenumber with the unit cm^-1 is rather used than the term frequency, however both magnitudes are propor-tional since the wavenumber is defined as the reciprocal of the wavelength. The output parameter usually generated by conventional FT-IR spectrometer (as is the case for the apparatus used in this work) is the absorbance A, defined as

⎟⎠

Characterization fundamentals 27 For the FT-IR analysis of thin solid films (such as PECVD a-Si1-xCx), infrared transparent substrates are needed. The most common choice here is float zone (Fz) silicon substrate due to its high purity, however in this work we also used Czochralski (Cz) germanium substrate for a more accurate analysis of the c-Ge/a-Si1-xCx system that was studied with respect to the surface passivation of germanium by PECVD films. The absorbance of the thin film Afilm is obtained by

sub total

film A A

A = − , _(3-3)

with A_total the absorbance of the coated substrate and A_sub the absorbance of the sub-strate only. Afilm is proportional to the absorption coefficient α of the film and is ob-tained by basically normalizing A_film with respect to the film thickness d

e d

A_film log10

= ⋅

α ^. _(3-4)

The IR transparency of the silicon and germanium substrate (Fig. 3-1) depends on phonon absorption, absorption due to impurities, plasma resonance and free carriers (doping level) and the surface condition (light scattering) [77]. For the evaluation of the film characteristics, it is therefore essential to use (almost) identical substrates.

The FT-IR measurements in this work were performed using an IFS-113v Fourier spectrometer from Bruker in the wavenumber range from 400 to 4000 cm^-1. Although the setup allows a resolution of down to 0.03 cm^-1, the resolution used in our experi-ments was set to 6 cm^-1, thereby reducing the influence of artifacts such as the so-called fringes. Details concerning the spectrometer as well as the origin and the elimi-nation of measurement artifacts can be found in [76].

Fig. 3-1: Absorbance of a p-type, 1.0±0.2 Ωcm, 250 µm thick, shiny etched Fz Si substrate (left) and of a p-type, 1.7±0.2 Ωcm, 500 µm thick, shiny etched Cz Ge substrate (right) measured by FT-IR.

28 Characterization fundamentals