We have measured the stresses in TaSi. films both as a function of annealing temperature and of composition of the as-deposited films. Substrates of Si, polysilicon and SiO2 were used. The residual, room temperature stresses in samples annealed at 900°C are noticeably dependent on the initial Si/Ta ratio. This effect appears to to be related to microstructural differences between Si-rich and Ta-rich films. The stresses are largely due to the mismatch of thermal expansion coefficients (TEC) between the silicide and silicon although there is also evidence that samples annealed at lower temperatures (500°C) may have a sizable component of intrinsic stress as well. Despite tensile stresses as high as 1500 MPa, the x-ray topographs of Si wafers with patterned silicide/polysilicon structures show that no substrate dislocations are present after- annealing.

Transport studies have been performed on thin films of CoSi 2 and NiSis2 in the temperature range 1 to 300 K. The conductivities are metallic with essentially the same temperature dependence; however, the residual resistivities are markedly different even though the two silicides are structurally similar (the room temperature resistivity of NiSi2 being at least twice that of CoSi2 of 15 μΩ cm). The difference is attributed to intrinsic defects in NiSi2. This defect has been simulated by ion bombardment of the film where it is also shown that Matthiesen's rule is obeyed over a remarkable range of bombardment doses.

Temperature-ramp Resistance Analysis to Characterize Electromigration (TRACE) has been applied to thin-film Al and Al-alloy conductors. Results have yielded activation energies in agreement with literature values. The TRACE technique has been used to determine the effect of H2 on the kinetics of electromigration damage (EMD) for both Al and Al-2%Cu thin-film conductors.

The reaction between cosputtered amorphous and polycrystalline tantalum silicide with Al-Si-metallization was investigated by means of Schottky and ohmic contacts on Si. Diffusson of Al and Si across the silicide was impeded up to 475 °C, as long as the silicide films contained excess Ta with respect to the TaSi2 stoichiometry. The results show that Al-Si with a thin Ta-rich tantalum silicide underlayer provides a VLSI-metallization with considerably reduced contact resistance, particularly for small contacts, and low Schottky barrier to n-Si without significant Si consumption from the substrate

The influence of deviations from TaSi2 stoichiometry in co-sputtered,amorphous Ta-Si films on interface reactions in contacts to silicon at 900°C was investigated by TEM of thin cross sections through the contact windows. Excess silicon precipitates epitaxially in the contacts, pits are formed for silicon deficiency. The involved lateral silicon transport ranges over several tens of microns. The effects of interfacial oxides and a crystalline TaSi 2 pad layer were studied.

The effect of excess silicon on the growth and microstructure of cosputtered TiSi 2 was investigated. Cosputtered films, with Si/Ti ratios between 2.5 and 4.5, were deposited on (100) Si substrates and reacted at temperatures from 650° to 1050°C. Excess Si in the asdeposited films formed a layer of epitaxial Si on the Si substrate. In addition, Si and SiO2 precipitates formed within the grains. The films became discontinuous when reacted at 1050°C.

Ion beam etching and reactive ion etching have been shown to significantly damage silicon surfaces. In this investigation, ion beam damage is examined as a function of ion beam type (1 keV Ne+, Ar+, and Kr+) and anneal history (temperatures of 500°, 800°, and 1000°C in ambient gases Ar,dN2, and Ar + H2 ). Damage at 1 keV, as characterized by currentvoltage and ESR measurements, appears to be roughly proportional to the inverse of ion size. Relatively long anneals at high temperatures (1 hr. at 800'-1000°C depending upon ambient) are found to be necessary to restore original surface characteristics.

The dry etching technologies reactive ion etching (RIE) and ion beam etching (IBE) have both been shown to cause a damaged layer at silicon surfaces. It has been demonstrated that this damage can be annealed out or, alternatively, it can be passivated with low energy hydrogen implants from a Kaufman ion source. This study further explores the hydrogen passivation approach by focusing on the effect of hydrogen implantation on damage caused by argon ion beam etching. The lighter hydrogen ions are actually shown ta cause more extensive damage than the heavier argon ions. However, by using low-energy hydrogen implants all damage, that present from the Ar and that generated during the hydrogen implant, can be passivated.

Defects introduced in p-type silicon during RF sputter deposition of Ti-W and electron-beam evaporation of hafnium were investigated using I-V, deep level transient spectroscopy and electron-beam induced current techniques. DLTS measurements indicate the presence of several deposition and evaporation induced defect states. H(0.35) at EV + .35 eV and H(0.38) were the most prominent defects. Minority carrier diffusion length results taken after annealing showed that in the case of the Hf contacts the damage was annealed out while in the case of Ti-W it was not. These differences in carrier recombination are traced to the concentration of H(0.35). Sputtering or evaporation induced damage also increased the barrier height. This observed increase was modeled assuming the introduction of donor-like defects.

Structures of the form Al-Bi-Se-metal were fabricated by deposition of the metals gold, cadmium, bismuth and thallium as thin films on a layer of crystallized selenium. The selenium was itself evaporated on a layer of bismuth on an aluminum base. This Al-Bi-Se base contact was rendered ohmic by substrate heating during the selenium deposition. The diodes showed forward-to-reverse current rectification ratios ranging from large for thallium to small for gold. At small forward voltage, the current density increased with increasing work function from thallium to gold, whereas the voltage intercepts Vo on plots of (capacitance)−2 versus reverse voltage decreased with increasing work function. From these Vo values and the slopes of the latter plots, barrier heights were deduced and plotted against work function. From this plot showing a decrease of barrier height with decreasing work function, it was possible to estimate an interface state density of 1013 cm−2 volt−1 assuming the existence of an interfacial layer 100Å in thickness. While the Se-Tl contact showed excellent rectification, it exhibited rapid degradation with time from atmospheric oxidation.

The boron minority carrier mirror (MCM) concept has been applied to Wacker poly-Si for passivation of active grain boundaries (GB). Photovoltaic, spectral response, laser beam induced current, electron beam induced current and interface state data show the MCM process to significantly improve solar cell performance. For example, an efficiency improvement from 6.4% to 8.8% was observed in part of the study. This appears to be a simple technique to produce stable solar cells with future application to fine grain poly-Si.

The use of spun-on liquids to form ohmic and Schottky barrier contacts on Si has been investigated. Two commercially available metallo-organic solutions containing Au or Pt were applied to Si substrates by spinning techniques. The Au or Pt contacts were then formed by annealing at 250°C or 450°C respectively. The metallurgical interactions between the spun-on Au or Pt layers and the Si substrate were investigated by MeV backscattering spectrometry and X-ray diffraction as a function of the annealing temperature. The resulting electrical characteristics were investigated with four point probe, I-V and C-V techniques. It was found that the spun-on Au or Pt films react with Si substrates at much higher temperatures than those deposited by vacuum evaporation. The diode characteristics of the spun-on Au films are comparable to those of evaporated Au films ( φB ∼0.85V, n ∼1.08), whereas diode characteristics of spunon Pt films show no linear region on the ln I-V curve. The application of spun-on Au also improves the bondability of Si chips on Mo headers.

A technique for electrostatic spray deposition (ESD) of thin films onto a conducting substrate is described. The method, using a 60 kV corona has been tested for the preparation of heat-reflecting tin oxide coatings onto window glass by spraying solutions of hydrated tin tetrachloride with 1% antimony trichloride in ethyl acetate.

Uniformity of film, growth rate, conductivity, infrared reflectivity and light transmission in the visible are all examined and compare favourably with films produced by normal aerosol spray. Conductivity and infra-red reflectivity increase with the substrate temperature during ESD. This is similar to the observation for antimony-doped tin oxide films prepared on soda-lime glass by CVD; although this temperature dependence does not follow closely that reported for normal spray deposition.

Depth profiles of boron implanted across the SiO2-Si interface have been measured by secondary ion mass spectrometry and compared with model calculations of the impurity distributions. Analysis of the sputtering and ion yield processes near the interface are used to reconstruct the measured boron profile for comparison with the model calculations. Applicability of the method of profile reconstruction near the interface to other materials and impurities is discussed.

Information relating the spatial arrangement and concentration of intentional and intrinsic dopants is commonly required to fully understand the properties of a material, whether the application is chemical, electrical, or physical. We have synergistically coupled the near-surface techniques of thermal neutron depth profiling (NDP) and secondary ion mass spectrometry (SIMS) for the purpose of better determining the distribution of a few key elements in a number of matrices and thin-film interfacial applications.

The NDP facility, unique in the U.S., allows virtually non-destructive measurements of the absolute concentration of specific elements (e.g,, He, Be, Li, B, Na, Bi . . .) to be made versus their depth distribution in a specific matrix [1]. The quantitative information is derived from the number and the residual energy of emitted charged particles that are produced in situ by uniformly illuminating a sample volume with thermalized neutrons. Sensitivity, depth of view, and resolution are dependent upon the reaction cross-section for the element of interest and the characteristic energy loss for the elemental components of the matrix. However, experimental parameters, such as the sample angle relative to the detector, can be adjusted to extract the maximum depth or the best resolution information from the measurement [2]. Since the technique is non-destructive, samples can be subjected to a series of treatments and profiled after each step [3].

The more mature SIMS technique is able to detect most of the elements listed above with greater relative sensitivity but without an absolute concentration calibration. Therefore, by utilizing the abundance information obtained by NDP, a concentration scale can be established for the SIMS profile. SIMS is also useful in probing smaller surface areas, a few tens of micrometers square as opposed to a few millimeters square for NDP. The advantage in coupling the two techniques lies principally with the role NDP plays in distinguishing experimental artifacts from real concentration variations [4]. While some matrices and interfacial areas of a sample give rise to variable sensitivities in SIMS measurements. NDP, however, counts every event that emitted a charged particle within the solid angle subtended by the detector, thereby, making it more reliable for reporting the concentration information.

Shown in Figure 1 is a comparison of NDP and SIMS profiles determined for a boron-10 implant in a single-crystal silicon, a common processing step for semiconductor materials. The agreement between techniques is good. Possible sources of discrepancies between the two methods are briefly discussed by Ehrstein et al. [3].

The combined effort of SIMS-NDP is currently being utilized to study diffusion and boundary segregation in thin-film semiconductor applications. Accurate depth profiles have been difficult to obtain by other analytical approaches for such material systems. The ability of SIMS-NDP to profile across interfacial regions and thin films will allow many other electrical devices and material problems to be addressed more reliably.

Quantitative depth profiles of impurities in LPE layers of HgCdTe have been determined using relative sensitivity factors calculated from ion implantation profiles. Standards were provided for Li, Be, B, C, F, Na, Mg, Al, Si, P, S, Cl, Cu, Ga, As, Br, and In. Relative sensitivity factors as a function of ionization potential for O2+ primary ion SIMS and electron affinity for Cs+ primary ion SIMS have been calculated in order to extend quantitation to elements not yet implanted. Examples of depth profiles for implant standards and unimplanted layers are given.

Schottky barrier structures often exhibit forward bias current voltage characteristics possessing a marked deviation from ideal behavior. In many cases, this deviation can be attributed to the influence of an appreciable series resistance on the diode characteristic making the determination of the actual Schottky barrier height from this characteristic difficult. In a study of rare earth silicides on Si, the present authors have successfully employed an alternative technique which overcomes this series resistance effect. This technique is derived from the conventional methods for obtaining barrier height and series resistance values in photovoltaic devices, although not necessarily requiring transparent metallizations. Photogenerated current from the device periphery is utilized here in the determination of the barrier height. The measurement of both open circuit voltage and the short circuit current at varying illumination intensities allows the junction characteristic to be determined essentially free from series resistance effects. A comparison will be made between the values of the barrier height derived from conventional J-V and photoresponse measurements, with photovoltaic measurements made on structures possessing either a semi-transparent or opaque metallization.