Reflectance Anisotropy Spectroscopy Spectra Research Articles

Growth phases and optical anisotropy of Co on preoxidized Cu(110)

We investigate the growth of Co on Cu(110) surfaces using oxygen as a surfactant. Reflectance anisotropy spectroscopy (RAS) and scanning tunneling microscopy (STM) were used for in situ analysis during different stages of growth. STM images taken after growth interruptions at various Co coverage exhibit a nearly layer-by-layer growth of up to 20 ML. For submonolayer coverage, different Co-Cu-O-phases are found, in agreement with previous work. Above monolayer coverage we observe a fairly smooth Co layer with an oxygen-terminated $\mathrm{Co}(110)(3\ifmmode\times\else\texttimes\fi{}1)$ surface structure. RAS spectra taken during growth interruptions reveal characteristic ``fingerprint''-like spectra assigned to the different growth phases. We interpret the features of the RAS spectra in terms of surface-state- and bulk-related electronic transitions. Moreover, we show that real-time monitoring of the growth process is possible by taking RAS transients at a fixed photon energy.

Reflection anisotropy spectroscopy: a new probe of metal surfaces

Reflection anisotropy spectroscopy (RAS) is a non-destructive optical probe of surfaces capable of operation within a wide range of environments. The RAS technique was developed to study semiconductor growth and more recently has been used to probe the optical properties of metal surfaces. Surface sensitivity has been demonstrated with the detection of reconstruction, adsorbate interaction and the observation of transitions involving surface-localized electronic states. The RAS technique has non-ultrahigh vacuum (UHV) environment capability and thus opens up the study of metal surfaces in catalytic and corrosive environments at high pressure and at the solid/liquid interface. The combination of RAS with a full range of surface science techniques applied to the study of well-defined surfaces aids in the interpretation of RAS spectra. The focus of this article will be on well-characterized metal surfaces prepared and studied under UHV conditions. Copyright © 2001 John Wiley & Sons, Ltd.

Atomic structure and optical anisotropy of III–V(001) surfaces

The optical anisotropy of materials with isotropic bulk crystal structure depends to a large extent on the surface atomic structure. For instance, data obtained by reflectance anisotropy spectroscopy (RAS) on (001) surfaces of zinc blende semiconductors such as InP and GaAs, have a fingerprint character for the various surface reconstructions. Here we present RAS spectra for GaAs(001) and InP(001) recorded at room temperature and at low temperature. We show that by comparison with a theoretical analysis based on ab initio density functional theory in local-density approximation calculations the origin of characteristic spectral features can be identified and thus RAS spectra utilized to discriminate between different competing structural models. We identify contributions related to electronic transitions between surface states as well as features arising from surface perturbed bulk wave functions. We explain the high sensitivity of RAS to the surface structure and chemistry as due to the surface state related features.

Development and Application of Instrumental Methods for Strain Analysisof Semiconductor Layers and Devices

Strain effects on semiconductor layers were studied by means of optical spectroscopic techniques with a device developed especially for the study of layered structures and microstructures. Raman, modulated photoreflectance and reflectance anisotropy spectroscopy (RAS) were applied. Measurements were performed on elemental semiconductors (Si), semiconductor alloys (Si–Ge) and III–V semiconductor compounds (GaAs). By application of RAS, strains lower than 10−4 could be resolved, which is at least one order of magnitude lower than those observable with Raman and modulated reflectance techniques. The RAS spectra of layers strained along either the [010] or [011] direction showed a derivative-like structure at E1-gap energies, which increased linearly and very quickly with increasing strain. The dependence of this spectral feature on applied strain was used to evaluate strain-dependent effects. This behaviour strongly suggests that RAS can be applied for the optical characterisation of strain in semiconductor microstructures and devices, with a higher efficiency and accuracy than that achieved by previously established optical methods such as Raman and modulation spectroscopy. In addition, the compactness and ease of operation of the instrumentation of RAS provides considerable potential for in situ monitoring/control of semiconductor fabrication conditions.

Optical characterization of indium-terminated GaAs(001) surfaces

We have investigated the growth of thin indium layers on As- and Ga-terminated GaAs(001) surfaces by reflectance anisotropy spectroscopy (RAS), low energy electron diffraction (LEED), and auger electron spectroscopy. Room temperature deposition of indium on the $(2\ifmmode\times\else\texttimes\fi{}4)/c(2\ifmmode\times\else\texttimes\fi{}8)$ surface and subsequent annealing at 450 \ifmmode^\circ\else\textdegree\fi{}C leads to the formation of an In-terminated surface showing a $(4\ifmmode\times\else\texttimes\fi{}2)$ LEED pattern, accompanied with strong changes in the measured surface optical anisotropy. When indium is deposited onto the $(4\ifmmode\times\else\texttimes\fi{}2)/c(8\ifmmode\times\else\texttimes\fi{}2)$ surface, on the contrary, the $(4\ifmmode\times\else\texttimes\fi{}2)$ In-terminated surface is already formed at room temperature deposition without needing annealing, as demonstrated by the RAS spectra. The finding that almost identical RAS spectra and $(4\ifmmode\times\else\texttimes\fi{}2)$ LEED patterns are obtained in both cases shows that the same final atomic structure is achieved. Finally, we conclude that the structure of the In-terminated surface is similar to that of the clean Ga-rich surface, although a more detailed model would need accurate calculations of the microscopic origin of the measured anisotropy.

MOVPE growth of (Al, Ga)InP-based laser structures monitored by real-time reflectance anisotropy spectroscopy

Using a specially-designed spectrometer enabling combined reflectance anisotropy spectroscopy (RAS) and reflectance measurements on rotating substrates in a commercial MOVPE reactor, we report the first full-spectroscopic RAS-monitoring of (Al,Ga)InP-based 650 nm laser growth. First, a spectral database was built up from systematic studies of AlGaInP RAS signatures for different Al compositions, doping levels and growth temperatures. These data are subsequently used for the interpretation of characteristic RAS fingerprints taken throughout the entire laser growth process. From the analysis of characteristic changes in the RAS spectra even small deviations from the optimum process (doping levels, composition, etc.) which would effect the performance of the final device can be detected.

Preparation of P-rich InP surfaces via MOCVD and surface characterization in UHV

For the first time direct contamination-free transfer to UHV was achieved for the P-rich InP(100) surface that is the easiest to prepare and control in the MOCVD environment. To avoid contamination during transfer a commercial MOCVD apparatus was modified to allow for transfer of samples to the 10-9 mbar UHV range within a very short time (less than 20 s) [1]. Epitaxial InP(100) films were prepared with TBP (tertiarybutylphosphine) and TMIn (trimethylindium) as precursors. In situ reflectance anisotropy spectroscopy (RAS) was carried out in the MOCVD environment. After transfer of the sample to UHV the same RAS spectrum was recovered. Auger-electron spectra (AES) confirmed the P-termination of the surface reconstructions suggested by RAS.

Growth of self-assembled (Zn)CdSe nanostructures on ZnSe by migration enhanced epitaxy

The growth of CdSe on ZnSe by migration enhanced epitaxy has been studied in situ by means of reflection high-energy electron diffraction (RHEED) and reflection anisotropy spectroscopy (RAS). The change from a streaky to a spotty RHEED pattern during CdSe deposition indicates the transition to a three-dimensional growth mode. This change is accompanied by a damping of RHEED intensity oscillations and a decrease in the overall intensity. RAS spectra show a pronounced change at a photon energy of about 2.2 eV with increasing CdSe thickness. To investigate the role of lattice mismatch, ternary ZnCdSe layers of different compositions have been grown. With decreasing Cd concentration the critical thickness for the growth mode transition increases and for Cd concentrations below 30% no growth mode transition could be observed.

Ge/GaAs(001) interface formation investigated by reflectance anisotropy spectroscopy

The formation of the Ge/GaAs(001) interface has been investigated following the transformation of an As-dimer terminated GaAs(001)(2 X 4) surface into a Ge-Ga-dimer terminated (1 X 2) reconstruction and the subsequent deposition up to 10 ML of Ge. The modification of the surface atomic geometry and the related electronic structure has been monitored by reflectance anisotropy spectroscopy (RAS) and low-energy electron diffraction. Experimental results are compared to density-functional-theory-local-density-approximation and tight-binding calculations of the surface structure and optical response, respectively. The comparison between calculated and measured RAS spectra allows us to show that the (2 X 4) structure transforms into a well-ordered (1 X 2) passing through a disordered (2 X 4) phase while a previously proposed intermediate (2 X 1) structure is ruled out. At higher Ge coverages, surface and Ge/GaAs-interface contributions to the optical spectra are separated by surface modification through exposure to atmosphere. A interface contribution is identified between 1.5 eV and 2.5 eV, almost identical in line shape and amplitude to the RAS features on the Ge-Ga-dimer terminated GaAs surface. This finding demonstrates that the backbonds of the Ge-Ga-dimers, present at the Ge-Ga-dimer terminated surface as well as at the Ge/GaAs interface, determine the optical anisotropy, whereas the Ge-Ga-dimer bond itself does not contribute significantly. [S0163-1829(99)12315-4].

Spectroscopic process sensors in MOVPE device production

Production environments require optical sensors more robust and compact than those used in research. In this paper such an optimized reflectance anisotropy spectroscopy (RAS) configuration is described and employed for MOVPE real-time growth monitoring. Measurements were performed in commercial epitaxial systems with rotating substrates. During growth of GaAs/InGaP heterojunction bipolar transistors (HBTs) RAS spectra were taken continuously with high time resolution. The signals are transformed into a color code for easy judgement during growth. Characteristic spectroscopic signatures can be used as fingerprints, providing information on surface stoichiometry and morphology, doping level, adsorbate coverage, layer relaxation and temperature changes at any stage of the growth process.

Raman and RAS Measurements on Uniaxially Strained Thin Semiconductor Layers

A sensitive and reliable method was developed for the determination of electron and phonon deformation potentials of thin layers and layered structures. A device was constructed and used to apply a line force at the center of a narrow stripe cut from a semiconductor wafer. The induced strain along the stripe was calculated from the maximum deflection at the center of the stripe. Thin layers of Si, SiGe and InGaAs were studied and the information on the phonon deformation potentials was deduced from the phonon frequency shift measured by Raman spectroscopy. Strain dependent changes in electronic band structure were monitored by means of Reflectance Anisotropy Spectroscopy (RAS). The RAS spectra consisted of sharp peaks at or near bandgap energies which increased rapidly with even a small increase of strain. The uniaxial strain device used with RAS is shown to provide a very sensitive technique for the detection of small amounts of strain. This method could be also applied for the study of superlattices.

Substrate dependence of adlayer optical response in reflectance anisotropy spectroscopy

Reflectance anisotropy spectroscopy (RAS) offers extreme surface sensitivity and can be used to study gas-solid and solid-liquid interfaces. Although the technique has already made a considerable contribution to in situ monitoring of the epitaxial growth of semiconductors, its impact has thus far been limited by difficulties in interpreting experimental results. In particular, RAS spectra are related to both Δε2′ and Δε2″, the real and imaginary parts of the surface dielectric anisotropy, with weighting factors determined by the complex dielectric function of the substrate. Here RAS calculations are performed for a variety of substrates assuming a model overlayer dielectric anisotropy. Three distinct regimes of behavior are observed: RAS spectra can resemble either Δε2′, Δε2″ or −Δε2″ (or some combination of the three) depending on the substrate and the photon energy. It is therefore crucial to properly account for these substrate effects if RAS is to be used to determine the azimuthal orientation of adsorbates or to study the symmetries of surface states, for example. Similar considerations also apply to the related technique of surface differential reflectance. The Fresnel-based theoretical method used here and the substrate optical functions presented should serve as a useful database for future work as RAS is applied to new systems.

Interpretation of reflectance anisotropy spectroscopy spectra of ZnSe(001) grown on GaAs(001) in terms of bulk, interface, and surface contributions

We combine reflectance anisotropy spectroscopy (RAS) with low energy electron diffraction, Auger electron spectroscopy, and soft x-ray photoemission spectra to analyze the surface optical anisotropy of ZnSe(001). Clean surfaces were prepared by thermal desorption of a protective Se cap from ZnSe layers of different thicknesses grown by molecular beam epitaxy on GaAs(001). Two surface reconstructions have been prepared by subsequent annealing, the Se-rich (2×1) reconstruction and the Zn-rich c(2×2) reconstruction. By modifying the surfaces either through submonolayer deposition of Sb or a short exposure to atmosphere it was possible to distinguish between surface and bulk/interface contributions to the optical anisotropy. Only on disordered, very Se-rich ZnSe(001) surfaces prepared at low annealing temperatures RAS features possibly related to electronic surface states are found. RAS spectra of the (2×1) and the c(2×2) surfaces are correlated with surface morphology and ordering rather than surface reconstruction and show features near the critical points of the bulk ZnSe band structure.

Reflectance anisotropy spectroscopy studies of the growth of carbon-doped GaAs by chemical beam epitaxy

Reflectance anisotropy spectroscopy (RAS) was used to study the carbon doping of GaAs grown by chemical beam epitaxy. Intrinsic carbon doping using trimethylgallium (TMGa) and arsine at a substrate temperature of 480°C was compared with extrinsic carbon doping using CBr4, triethylgallium and arsine. Samples were grown under a range of V : III ratios and the resultant surface reconstructions were determined by RAS. For samples grown under a (2×4)-like reconstruction the RAS spectra taken immediately after growth showed a strong (2×4)-like character rather than a return to a c(4×4)-like reconstruction as is seen for undoped GaAs. This was seen for both intrinsic and extrinsic doping and is attributed to a shift in the As-dimer bond energy for highly doped GaAs : C. Features in the RAS spectra at 3.1 eV which developed during the cooling of the samples and which were clearly related to the carbon doping level were attributed to the linear electro-optic effect caused by doping of the underlying layer and not to surface structural effects. The C-doping level obtained using TMGa is known to vary strongly with V : III ratio and this was confirmed by post-growth analysis. IR spectroscopy of heavily doped samples grown using TMGa showed the presence of an aligned carbon complex (H–(CAs) 2), lying parallel to [1 1 0] which has previously been seen in GaAs grown by metal-organic molecular beam epitaxy (MOMBE).

Ellipsometric and reflectance-anisotropy measurements on rotating samples

Device grade semiconductor epitaxial growth is usually performed on rotating substrates in order to achieve and composition uniform layer thickness. This often complicates in-situ optical measurements for growth control. Therefore, in this paper we present a method that significantly reduces the signal distortions imposed on RAS (reflectance anisotropy spectroscopy) and ellipsometry spectra by rotating and wobbling samples. Several cases (phase and frequency of the sample rotation either known or unknown, wobbling and ideally rotating samples) are investigated and the respective optimized experimental set-ups are presented for both RAS and ellipsometry.

On-line growth monitoring of InP-based device structures by reflectance anisotropy spectroscopy

Reflectance anisotropy spectroscopy (RAS) has been used to study the metalorganic vapor phase epitaxy growth process for GaxIn1−xAsyP1−y/InP light emitting diodes. The sensitivity of RAS to morphology changes is demonstrated by InP growth on different InP:Fe substrates. RAS reveals not only development of dull surfaces but also detects initial temporary roughness of mirror-like layers. Based on the RAS results the substrate preparation was optimized. RAS spectra measured on n- and p-type InP and p-type GaInAsP during light emitting diodes production are suitable for finger-printing of the growth process. Spectra from InP:Si and InP:Zn layers show characteristic features near 4.3 eV which allow for assessment of doping level at growth temperature (640°C). Correlation of RAS spectra and transients during growth with the quaternary composition was achieved. A change in composition of only Δx=0.01, Δy=0.03 corresponding to a shift of photoluminescence-peak position by 16 nm was detectable in RAS spectra. The results demonstrate the high sensitivity and thus the suitability of RAS for on-line control during growth of device structures.

Formation of GaAsP interface layers monitored by reflectance anisotropy spectroscopy

Reflectance anisotropy spectroscopy (RAS) has been used to study As-by-P exchange during metalorganic vapor phase epitaxy. The study focuses on the processes occurring during switching from GaAs to GaInP, especially the effect of purging PH3 over a GaAs surface. GaAsP/GaAs superlattices of different periodicity were grown and the P-content was determined by high-resolution x-ray diffraction and correlated to the RAS spectra. From the temperature dependence of the P-content, an activation energy of 0.56 eV was estimated for the incorporation mechanism. In addition to the insights into the processes at mixed group-V heterointerfaces, our study demonstrates the reproducibility of RAS transients that thus can be used for process monitoring.

Structure of InP (001) surfaces prepared by decapping and by ion bombardment and annealing

The structure of InP surfaces prepared by ion bombardment and annealing (IBA) and by decapping of InP (001) samples grown by metal organic vapor phase epitaxy is studied. The structural changes of the surfaces during preparation are monitored by low-energy electron diffraction and related to the surface electronic modifications as revealed by reflectance-anisotropy spectroscopy (RAS). For both preparation methods we find (2\ifmmode\times\else\texttimes\fi{}4) reconstructions and almost identical RAS spectra. This finding contrasts with previous reports usually claiming a (4\ifmmode\times\else\texttimes\fi{}2) surface after IBA Auger-electron spectroscopy as well as hydrogen-adsorbate vibrations recorded with high-resolution electron-energy spectroscopy indicate an In-rich surface stoichiometry.

In situ photoemission and reflectance anisotropy spectroscopy studies of CdS grown on InP(001)

Structural and electronic properties of the clean InP(001) surface and the CdS/InP(001) interface were investigated by soft x-ray photoemission and reflectance anisotropy spectroscopy (RAS). Clean InP(001) (2×4) surfaces were prepared by desorption of a protective arsenic/phosphorus double capping layer, thereafter, CdS deposited by single source molecular beam epitaxy at a substrate temperature of 200 °C. Core-level and valence-band photoemission spectra, as well as the RAS data, were taken in situ. Core-level spectra show a disappearance of In 4d and P 2p surface core-level shifts and a chemical reaction between indium and sulphur upon CdS deposition. From valence-band spectra, a total valence-band offset of ΔEV=(0.75±0.10) eV is derived, which corresponds to previous results of the CdS/InP(110) interface. After deposition of thin CdS layers, RAS spectra show strong features close to interband critical points of bulk InP due to surface roughening induced by the interface reaction. A feature around 5.3 eV, developing with higher coverage, indicates the formation of the metastable cubic phase of CdS as confirmed by additional Raman spectra.

Theoretical study of the Si/GaAs(001)-c(4×4) surface

The deposition of Si on the GaAs(001)-c(4×4) reconstructed surface is studied with the aid of ab initio pseudopotential calculations. With reference to the results of scanning tunneling microscopy (STM), reflectance anisotropy spectroscopy (RAS), and reflection high-energy electron diffraction experiments, models for the Si/GaAs(001)-c(4×4) surface are constructed with Si coverages corresponding to 0.25 and 1.0 ML. STM images are simulated and RAS spectra are calculated for these models and compared with experiment. By examining the local density of states and the magnitude of the optical transition matrix elements it is possible to make a link between the features in the RAS spectra and surface electronic structure. Comments are made on the validity of these surface structural models