We have developed a novel stand-alone diagnostic system that can analyze a semiconductor wafer surface in each process without introducing contamination. This allows us to analyze the relationship between chemical conditions and device properties.

A UHV (Ultra High Vacuum) wafer transfer chamber is used between the measuring apparatus and the semiconductor processes. The chamber vacuum system, which consists of a battery driven ion pump and a liquid N2 shroud, achieves a pressure of 2 × 10−8 Pa (corresponding to about 100 min. until one monolayer of contamination has been adsorbed).

Wafer transfer lines have been constructed between semiconductor vacuum processes, CVD (Chemical Vapor Deposition) and measuring instruments, ESCA (Electron Spectroscopy for Chemical Analysis) and TEM (Transmission Electron Microscope). Our results from ESCA and TEM showed measurements that carbon contamination and oxidation was suppressed.

We have developed a technique which enables us to reduce the self-bias voltage without introducing any substantial disturbance to the main plasma. This approach is based on decreasing the floating potential of the powered electrode by means of injecting an electron current from the electrode. Instead of employing a hot cathode, we have used an auxiliary glow discharge to supply the electron emission. The controllable range of the self-bias has been extended to almost 0 V. A theoretical explanation of the variation of the floating potential with the emission current has been carried out. Measurements of the floating voltage(Vf) are plotted against ln(1/Iau), where Iau is the auxiliary discharge current. A good linear relationship of Vf vs. ln(1/Iau) is obtained as expected from the theory, and its slope gives the electron temperature.

In order to appraise the degree of damage induced by the bombardment of ions accelerated by the sheath voltage, we have estimated the damage by measuring the dark current of a sample diode fabricated from the film. It has been observed experimentally that when the sheath voltage was depressed from -73 V to -36 V at a pressure of 0.11 Torr, the dark current of the sample diode was suppressed drastically from 1.3 ×10−10 A to 5.3×10−11 A

In this work, Langmuir probe measurements were used to characterize a multipolar electron cyclotron resonance (ECR) plasma source. This system has many controllable parameters including microwave power, rf power, gas, pressure, flow rate, and source distance. Both double and triple Langmuir probes were used for the plasma characterization. The results from the Langmuir probe measurements were correlated to the etch characteristics of photoresist. Ion density and photoresist etch rate were found to increase with microwave power but decrease with source distance. However, rf power does not have significant influence on ion density although the photoresist etch rate increases substantially with if power. Ion density first increases then decreases at higher pressure. Maximum ion density occurs at lower pressure for larger distance below the ECR source. Ion density uniformity for an O2 plasma is ±2% across a 16 cm diameter region at 23 cm below the source. For photoresist etched at 10 cm source distance, etch rate uniformity is ±2% for a 15 cm diameter wafer. The results from the Langmuir probe measurements indicate that photoresist etching is enhanced by ion density and ion energy.

Reactive ion etching of silicon and germanium in CF4 -O2 was investigated. Above 20% O2 germanium etching is selective with respect to silicon. In agreement with the evolution of the fluorine and oxygen concentration in the plasma and of the etch products formation rate, surface analysis reveals that the growth of a SiOxFy layer slows down the silicon etching whereas the formation of GeOxFy does not inhibit germanium etching. Using a simple kinetic model, the silicon and germanium reactivity and its dependency with respect to the plasma composition are expressed in function of the experimental data. Results suggest that surface composition controls silicon etching, whereas germanium etching depends only on the fluorine flux on the surface.

We describe an experimental method capable of evaluating the elastic and loss moduli and the thermal diffusivity in thin films. The versatility of the technique is demonstrated with data scans from several films of technological interest.

The advantages of in-situ SIMS plasma probe diagnostics are highlighted in low pressure hydrocarbon ECR reactive ion etching (RIE) of III-V materials. Three aspects of the RIE process are investigated. First, the dominant ion species in a CH4/H2/Ar plasma are recorded at various chamber pressures, ECR powers, CH4/(CH4+H2) gas flow ratios and microwave cavity tuning. These studies have improved our understanding of the effects of these parameters on the relative concentrations of reactive precursor species in the plasma and have led to more rapid optimization of the etch system. Secondly, SIMS has been used for identification of reaction products from the III-V surface at the optimized plasma conditions. The Ar diluted mixture gives rise to significant levels of group V hydrides and organometallic compounds and the dominant group III volatile ions have been positively identified as dimethyl species. The third and final aspect reported is the application of volatile product identification to endpoint detection. In lcm2 multiple quantum well samples, layers as thin as 50Å are easily distinguishable.

Quadrupole mass spectrometry (QMS) has been used as an in-situ diagnostic technique for GaAs etched with an electron cyclotron resonance source. Changes in the detected signal intensities for reactive species and etch products have been related to corresponding changes in the etch rate as several process parameters were varied. The detected 75As+ and to a lesser degree, 35C1+ and 70C12+, were observed to follow etch rate as microwave power, rf power, source to sample distance, temperature, and pressure were varied. The self-induced dc bias (IVdcl) determines the etch rate dependence on etch time. The time delay before saturation of the monitored 75As+ signal corresponding to a constant etch rate is inversely proportional to IVdcl. The addition of N2/O2 in a 4:1 ratio to constitute 15% of the total discharge resulted in a 95% decrease in the intensity of the monitored 75As+ signal. The measured etch rate decreased by 75%.

The case for incorporating an arsenic capping layer in compound semiconductor device processing has been investigated with x-ray photoelectron spectroscopy (XPS), low-energy electron diffraction (LEED) and scanning electron microscopy (SEM). The As cap was found to be stable upon exposure to common processing chemicals, such as acetone, photoresist, developer, and N-methyl-2-pyrrolidone, a common polyimide solvent. A clean, c(4×4)-reconstructed GaAs(001) surface was recovered after thermal desorption of the cap in ultra-high vacuum, for a sample exposed to standard (maskless) photolithography. We also report a new technique for reactive decapping at room temperature, using a beam of hydrogen radicals (H*). Pattern definition in the As cap with ∼ 5 μm linewidth was demonstrated, using this technique. However, XPS and SEM data for the H*-etched specimens showed clear evidence of superficial gallium (sub)oxide and of As residues along the photoresist mask edges. This novel method of As cap patterning thus needs further refinement, before being useful to III-V device processing.

Narrow (∼30 μm ϕ) via holes have been etched in both InP and GaAs substrates using either C12/CH4/H2 /Ar or BC13/C12 discharges, respectively. High density (∼5×1011 cm−3), low pressure (1 mTorr for C12/CH4/H2/Ar or 15 mTorr for BC13/C12) conditions, combined with sidewall passivation obtained using AZ 4620 photoresist masks, produce the correct profiles for subsequent metallization to complete the via connection. Optical emission monitoring of the 417.2 nm Ga line during GaAs etching or of the 325.6 nm In line during InP etching provided a sensitive, non invasive and reliable indicator of endpoint for both types of substrates. The intensity of these lines was proportional to the microwave input power at fixed dc bias and pressure. The via holes are suitable for a range of InP and GaAs microwave power devices, including Heterojunction Bipolar Transistors and High Electron Mobility Transistors.

The deposition of thin films by magnetron sputtering is readily influenced by the reduction/oxidation characteristics of the plasma. Unfortunately, the redox state of the sputter plasma is often inadvertently determined by uncontrolled factors such as transient evolution of species from the target. Undesirable variations in film and target properties can be avoided by actively setting the redox conditions. This is most readily accomplished by including two competing species, one oxidizing and the other reducing, in the chamber gas mixture. Optical emission spectroscopy, an in-situ process monitor, was employed to observe redox interactions in nominally unreactive dc sputtering of YBCO and reactive rf sputtering of IrO2. Optical spectroscopy of YBCO sputtering reveals that the intensity of atomic oxygen emissions decreases with H2 additions to the sputter gas blend.

The use of band-edge reflection spectroscopy (BRS) to determine the substrate temperature during MBE is reviewed. Data are presented for Si, GaAs, InP and CdZnTe substrates, and the use of BRS during the growth of ZnTe on Si is demonstrated. We discuss complications that arise due to optical interference in the epitaxial layers, and methods to compensate for the effects of interference are described.

Single photon laser ionization time-of-flight mass spectroscopy (SPI-TOFMS) is used to monitor the gaseous fluxes of Ga and Asn, during molecular beam epitaxy of GaAs. This noninvasive and real-time probe measures densities, and hence fluxes, of multiple chemical species impinging on or scattered from a substrate during conventional MBE. With single photon ionization at 118 nm (10.5 eV, ninth harmonic of Nd:YAG laser), the photon energy is large enough to ionize the species, but insufficient to both ionize and fragment. The lack of molecular dissociation of As2 and As4 greatly simplifies the interpretation of mass spectra. Additionally, the geometry of the single photon ionization TOFMS permits simultaneous film growth monitoring using RHEED. Results will be presented on the probing of scattering and desorption of III-V MBE species during GaAs growth. This technique promises to be a valuable in-situ diagnostic for III-V and II-VI MBE.

Real-time epi-film thickness is measured by an Emission Fourier Transform Infrared Spectrometer (E/FT-IR). The E/FT-IR takes advantage of the heated wafer as the source of IR radiation. In our experiments, wafers were cleaned using in-situ ECR hydrogen plasma followed by film growth. The cleaning and deposition processes were monitored in real-time using the E/FT-IR technique. We have demonstrated the application of E/FT-IR for observing real-time growth rates and incubation times. Based on these real-time observations, the predeposition plasma cleaning process and the deposition process can be effectively monitored and controlled in real-time. Application of E/FT-IR in optimizing the predeposition hydrogen plasma cleaning process was demonstrated.

Electron traps in Czochralski–grown n-type (100) silicon with and without donor annihilation annealing have been studied by deep–level transient spectroscopy. A total of eight electron traps are observed in the concentration range 1010 –1011 cm −3. It is thought that these are grown–in defects during crystal growth cooling period including donor annihilation annealing. It is suggested that two electron traps labelled A2 (Ec–0.34 eV) and A3 (Ec–0.38 eV) of these traps are correlated with oxygen–related defects. It is shown that traps A2 and A3 are formed around 400 ° C and disappear around 500–600 ° C.

The relationship between minority carrier properties and solar cell characteristics of electromagnetic (EM) cast polycrystalline Si has experimentally been investigated. The minority carrier lifetime τ and diffusion coefficient D were evaluated by a novel dual mercury probe method. The solar cell characteristics, e.g., a conversion efficiency η were measured by fabricating experimental solar cells using the corresponding wafers. The wafer showing high-η (13.1%) has relatively high τ (av. 8.2 μs) with small variation of I) (av. 29.6 cm2/s). On the contrary, the low-η (11%) wafer shows low τ (av. 1.1 μs), including some inferior portions with very low τ of less than 0.5 μs. It is also shown that D drastically deteriorates with decreasing τ if τ is less than around 2 μs. To realize high efficiency polycrystalline solar cells, the wafers with high value of τ and without considerably low-τ portions are needed.

Many recent studies of the highlighted problem of visible-range photoluminescence of porous silicon (po-Si) indicate the presence of hydrogen in this material. However, its actual role is not clear with the discussed possibilities ranging from passivation of the surface defects to some form of active participation in the photoluminescence (PL) mechanism itself. At the same time, in several magnetic resonance studies of po-Si the so-called Pb center, originally identified with the <111>-directed silicon broken bond stabilized at Si/SiO2 interfaces, has been reported.

Very recently the paramagnetic state of molecular hydrogen in bulk silicon has been identified. The spin-Hamiltonian parameters of the related Si-NL52 EPR spectrum appear to be identical to those of the Pb center. This observation casts severe doubts on the original interpretation of the Pb center while, on the other hand, offering new evidence of a prominent presence of hydrogen molecules both in po-Si and at the Si/SiO2 interface. In this paper the similarities of Si-NL52 and Pb centers are examined in detail, and it is argued that the centers are the same. It is concluded that many features of the Pb center which could not be explained within the broken-bond model can be understood with the interstitial hydrogen molecule model.

Progress is reported in developing reliable methodology for imaging silicon surfaces with the atomic force microscope (AFM). A new form of AFM, known as tapping mode AFM, has been found to provide the best quality data for surface roughness determinations. Commercially available colloidal gold spheres have been used to fabricate tip characterization standards and are used to report tip size with roughness data. Power spectral density calculations are shown to provide a useful roughness calculation based on lateral wavelength.

We have characterized the surface topography of silicon films from different deposition and doping process sequences using AFM and optical reflectivity. The resulting surface structures after deposition, doping, oxide growth, and oxide removal correlate with the electrical leakage currents and breakdown voltages of double polysilicon capacitors. As-deposited amorphous films had smoother surfaces than those deposited in the crystalline state. Gas-phase diffusion doping increases the surface roughness. Only the amorphous in situ doped films retained a smooth surface following oxidation, yielding low leakage capacitors with breakdown fields above 8 MV/cm. Surprisingly, implanted amorphous films exhibited the roughest interfaces, resulting in lower breakdown fields. This study has shown that AFM provides an effective, quick, non-destructive diagnostic technique for semiconductor processing.

A reliable nondestructive method for measuring trace amounts of hydrogen in semiconductors and related materials has long been needed. Cold neutron capture prompt γ-ray activation analysis (CNPGAA) is a nondestructive, multielement technique which has found application in the measurement of trace amounts of hydrogen. The sample is irradiated by a beam of “cold” neutrons; the presence of hydrogen is confirmed by the detection of a 2223 keV gamma-ray. The technique gives bulk analyses (the neutron and gamma radiation penetrate the sample), the hydrogen peak is free of interferences, and the results are independent of the chemical form of hydrogen present. The instrument is capable of detecting less than 10 mg/kg of hydrogen in many matrices. We have used the technique to measure hydrogen levels in a dielectric film on a silicon wafer, semiconductor grade germanium, and quartz.

Medium-energy ion spectroscopy using a conventional ion implanter has been developed to study the properties of semiconductor subsurface regions. The system is equipped with solid state detector and operaties with He+ ion energy up to 200 keV. We have tested the system performance for various applications, such as, silicon diffusion through a thin Au layer, a low dose, low energy As implantion and damage of silicon surface caused by plasma treatment.