The time/temperature dependencies of the surface roughness, subsurface extended defect formation, and minority-carrier lifetime are reported for n-type (100) silicon wafers exposed to a hydrogen ion beam. Surface roughness is assessed from atomic force microscopy, the distribution and nature of extended defects are determined from transmission electron microscopy, and the minority-carrier lifetime is evaluated by a non-contact laser-microwave technique. The surface roughness exhibits a weak dependence on ion-beam exposure time for the temperature range studied, whereas the distribution of extended defects may depend on exposure time at a given wafer temperature. The surface and bulk components of the minority-carrier lifetimes are consistent with these surface and subsurface properties. Transmission electron microscopy analyses demonstrate that the associated strain field of the extended defects is compressive in nature.

Theoretical basis of Photoconductance Decay and Surface Photovoltage Techniques is presented. Impact of typical measurement set-ups and sample preparation procedures on the differences observed in experimental results which include diffusion length, recombination lifetime, and bulk iron concentration is emphasized. Measurement accuracy and evaluation repeatability for both techniques is simulated. Relevant experimental evidence is presented based on a variety of samples with different intentionally introduced metal contamination, oxygen levels, and surface electrical properties.

Photoluminescence (PL) studies were applied to intentionally iron contaminated silicon wafers to validate the PL technique for the quantitative evaluation of bulk iron in silicon. iron contamination ranged from 10(9) cm(-3) to 10(12) cm(-3). For lightly doped p-type and n-type silicon a good correlation was found between Photoconductance Decay (PCD) lifetime, Surface Photo Voltage (SPV) diffusion length (DL) and iron readings and PL intensity readings. PCD, SPV and PL high-resolution mapping was applied to build the point-to-point correlation. PL was shown to be sensitive to iron contamination at concentrations exceeding 10(10) cm(-3) as calibrated by SPV using lightly doped p-type silicon. PL may be used as high-resolution non-destructive technique to track down the metal contamination sources in wafers processing.

We analyze the structure and defect formation of heavily As-doped Czochralski (CZ) grown Si crystals with resistivities below 3 milli-ohm cm based on results obtained by x-ray diffraction, photoluminescence and optical microscopy. The lateral interference fringes on the sectional x-ray topographs of heavily As-doped Si crystals was found to disappear due to crystalline lattice distortions that is most likely associated with As clustering. While arsenic doping of CZ Si at levels of 4x10(19) atoms x cm(-3) May lead in As clustering, it does not necessarily result in structure loss. Dislocation generation is mainly a function of crystal growth conditions that may lead to the corrugation of the crystal-melt interface, columnar growth and structure loss.

A photoluminescence (PL) signal has been modeled under conditions of steady-state, depth-dependent excitation, ignored surface recombination, and neglected carrier diffusion. In spite of serious model limitations, the experimental data of PL vs. resistivity collected in the 0.0009 to 20 Omega cm range showed a reasonably good agreement with the simulation output. The PL signal reaches maximum within the 0.01 to 0.1 Omega cm range. At lower resistivity, the signal strongly depends on doping level and it can be used for doping striation monitoring. At higher resistivity, PL depends on the applied excitation levels and the concentration of deep recombination centers (material contamination). Within the entire range, PL can be used for imaging of extended defects, such as dislocations and precipitates. (C) 2003 The Electrochemical Society.

PL studies of oxygen precipitation related defects, stress relaxation related defects and doping striations in various silicon materials are presented. The sample spectrum includes a variety of dopant species, and the dopant concentration range covers several 10 14 cm −3 to several 10 19 cm −3 . Lightly doped, precipitation-annealed polished wafers were intentionally contaminated with Fe, Ni and Cu. Several types of epi wafers based on heavily doped substrates have been investigated after full device processing. PL intensity in the investigated doping concentration range is controlled by three basic recombination mechanisms: radiative recombination competing with multi phonon Shockley-Read-Hall (SRH) and Auger recombination. SRH recombination is the major competing mechanism at low dopant concentration, and Auger recombination becomes important at increasing doping levels. Even though not yet fully understood, the PL technique applied in this study has generated practically useful results.

Surface photovoltage (SPV) measurements are traditionally carried out under steady-state conditions to determine the minority carrier diffusion length. While this technique is very convenient for bulk wafer defect characterization, especially the detection of iron in boron-doped silicon wafers, it is poorly suited to characterize epitaxial layers that are typically much thinner than the minority carrier diffusion length. We have developed the theory for frequency-dependent SPV measurements and have verified this theory with experimental data. We consider the various recombination/generation components in the semiconductor and determine the dependence on photon flux density, optical absorption coefficient, doping density, recombination lifetime, and temperature. Epitaxial layers are usually measured with techniques that are sensitive to generation parameters confined to the reverse-biased space-charge region (scr). We show that optical excitation can be used for scr confined recombination measurements, but the resultant lifetime is an effective lifetime incorporating both scr and surface recombination, heavily influenced by surface recombination. (C) 2001 The Electrochemical Society. [DOI: 10.1149/1.1380257] All rights reserved.

Surface photovoltage (SPV) measurements are traditionally carried out under steady-state conditions to determine the minority carrier diffusion length. While this technique is very convenient for bulk wafer defect characterization, especially the detection of iron in boron-doped silicon wafers, it is poorly suited to characterize epitaxial layers that are typically much thinner than the diffusion length. We have developed the theory for frequency-dependent SPV measurements and have verified this theory with experimental data. We consider the various recombination/generation components in the semiconductor sample and determine the dependence on photon flux density, optical absorption coefficient, doping density, recombination and generation lifetime, and temperature. We show that optical excitation can be used for space-charge region recombination measurements, provided the optical generation rate is lower than the thermal generation rate thereby making such measurements useful for epitaxial layer characterization.

Recombination lifetime measurements were applied for assessing the gettering strength of lightly and heavily boron-doped 200 mm CZ-Si wafers. Photoconductance decay (PCD) is shown to reveal lateral gettering variation correlated with inhomogeneous bulk defect generation in lightly doped wafers. Injection level spectroscopy allows for differentiation of non-gettered iron and bulk defects. Photoluminescence in mapping mode (PLM) gives an equivalent information for heavily doped material, High spatial resolution allows for single bulk defect analysis. From the recombination activity of bulk defects we conclude on gettering phenomena.

Applicability of a new contact-less metrology used for evaluation of near-surface doping and lifetime in thin epi layers or denuded zones, based on frequency resolved surface photo voltage, has been studied in this work. The metrology is facilitated by recently introduced commercial instrument called Epi-tau, manufactured by Semiconductor Diagnostics, Inc. Simple modeling shows that the measured time constant is not related to the recombination lifetime but rather surface barrier height, recombination velocity and doping. The barrier height and recombination velocity are primarily dependent on the surface passivation state (oxidation). The experimental data indicates that the spectra are also dependent on bulk recombination lifetime and substrate properties, however a more comprehensive model is needed for better data interpretation. In spite of some deficiency in the comprehension of the new approach, the instrument can deliver valuable practical data on doping and lifetime useful in epi or denuded zone process monitoring.

COCOS (Corona Oxide Characterization of Semiconductor) is a noncontact diagnostic technique in many respects superior to MOS methodology In this technique, an electric field in the oxide and in the semiconductor is changed by dosing an electric charge on the oxide using a corona discharge in air. The response is measured by the contact potential difference in the dark and under strong illumination giving the oxide voltage, V-OX and the surface barrier, V-SB, and in turn the oxide capacitance, C-OX the flat band voltage, V-FB, and also interface trap spectra, D-it across the silicon bandgap. In thin gate oxides, COCOS provides a monitoring technique for oxide reliability. The corona charging is used to stress the oxide and monitor COCOS stress-induced wearout through the measurement of D-it and V-FB Shift. The newly introduced COCOS I-V characteristic is used to monitor stress-induced leakage current, SILC. COCOS-SILC reliability results are provided showing the sensitivity of this approach to GOI defects originating from heavy metals, COPs (crystal originated particulates) and surface roughness in gate oxides fi om 3 to 20 nm in thickness. In view of the present findings it is apparent that the magnitude of COCOS-SILC at a constant oxide field can serve as a sensor of GOI defects. It was also established that without corona stress the tunneling current is practically insensitive to GOI defects.

The effect of the carrier recombination process in silicon on the microwave reflection coefficient is analyzed in the frequency domain. The process is described using a two level recombination/trapping model. Carrier recombination kinetics are characterized by four parameters, two of which are related to the recombination and the other to the trapping processes. These parameters are evaluated for Czochralski silicon wafers based on Nyquist plots. In the evaluation procedure, a nonlinear simplex method is used for fitting the experimental data to the model.

The temperature dependence of the minority carrier lifetime and the surface recombination velocity have been measured in p-type Si wafers containing 10(11) cm(-3) Fe, using Frequency Resolved Photoconductance (FR-PC) method. The recombination trapping centers have been identified from the microwave signals drawn as Nyquist plots. Changing the temperature and de injection level enables one to characterize the traps with the nonlinear simplex fitting procedure. Under moderate de injection level the Nyquist plots exhibit a large are over the temperature range 24 degrees C to 240 degrees C. For low de injection, and at temperatures below 100 degrees C, the principal are is deformed giving rise to second are. However, this deformation and the second are disappear above 180 degrees C. Optimum fitting procedures show that the principal are is associated with carrier recombination due to interstitial Fe-i, while the are deformation is attributed to electron trapping at this level. The second are can nor be fitted by the above models; however, the Lorentzian amplitude and peak dependence on de injection level leads us to believe that another level, donor Fe-B, is involved in electron trapping. Observed discrepancy at the room temperature between the effective lifetime and the frequency resolved lifetime is attributed to this level. Above 180 degrees C both lifetimes correlate perfectly.

In this work the impact of metals such as Cu, Fe, Ti, Ni, Cr, Mn, and Zn intentionally introduced to n- and p-type silicon crystals during crystal growth on recombination lifetime evaluated by Surface Photovoltage and Photoconductance Decay techniques is presented. Metals-defects interactions resulting in changes of lifetime absolute values and its radial distribution, due to the thermal oxidation process inherent to common PCD preparation step, are discussed.

Gate oxide integrity (GOI) tests, surface photovoltage and deep level transient spectroscopy of Czochralski silicon wafers reveal oxide degradation at heavy precipitation, defect-controlled recombination lifetime and defect-induced deep levels. Electron beam induced current measurements on those wafers before and after intentional metal decoration reveal relatively shallow levels in the non-decorated state and deep levels in the decorated state. It is shown that the actual contamination level determines the usefulness of GOI tests for predicting material performance in device processing.

A laser microwave photo conductance decay (LM‐PCD) frequency resolved method has been used for determining charge carrier trapping parameters. A linear theory is used to calculate the output microwave detector signal as a function of a frequency. Fitting the theoretical analysis to experimental data for a single FeB level has been achieved for Fe contaminated p‐type Si.