Bombardment of slightly rough, planar surfaces with an energetic gas-cluster ion beam (GCIB) is known to result in a decrease in roughness. We report comparison of measured surface evolution with simulations based on a simple phenomenological model. The model consists of a continuum rate equation for the surface height undergoing multiple cluster impacts. With suitable assumptions the main features observed in the evolution of an initially nonplanar surface can be simulated. Qualitatively, both experiment and simulation show that sharp features and asperities are rapidly eroded. Examples are given of initial surfaces that are either randomly rough and scratched, or very smooth with added nanoscale engineered features. Both types of surfaces show smoothing in microscopy evaluations of the GCIB and in the simulations, thereby identifying impact-transient surface diffusion as the primary cluster smoothing mechanism.

Thin-film magnetic sensor and memory devices in future generations may benefit from a processing tool for final-step etching and smoothing of surfaces to nearly an atomic scale. Gas-cluster ion-beam (GCIB) systems make possible improved surface sputtering and processing for many types of materials. We propose application of GCIB processing as a key smoothing step in thin-film magnetic-materials technology, especially spin-valve GMR. Results of argon GCIB etching and smoothing of surfaces of alumina, silicon, permalloy and tantalum films are reported. No accumulating roughness or damage is observed. The distinct scratches and tracks seen in atomic-force microscopy of CMP-processed surfaces, are removed almost entirely by subsequent GCIB processing. The technique primarily reduces high spatial-frequency roughness and renders the topographic surface elevations more nearly gaussian (randomly distributed).

Surface processing of microelectronic materials by bombardment with nanoparticles of condensed gases (i.e., clusters) in the form of an ion beam, makes possible etching and smoothing of those surfaces to very high figures of merit. As this is not possible with any conventional ion method, gas-cluster ion-beam systems have great potential in manufacturing. The formation of gas clusters and their collision with surfaces provides an interesting arena for novel physics and surface science. This paper outlines a physical model for the clusters and surface interactions, and provides examples of surface processing. In particular, the reduction of surface roughness while etching by cluster-ion bombardment is illustrated for various materials utilized in microelectronics.

The surface morphology of 40-MHz PECVD SiNx films is investigated. We report on the correlation between the deposition conditions, bulk properties, and surface roughness of these TFT insulators. The roughness is measured with atomic-force microscopy, AFM. A link will be presented between the AFM properties and the effects of hydrogen dilution during deposition: gas composition and rf-power-density (P) dependences will be discussed. An increase of the surface roughness to 3.7 nm is observed upon H2 dilution and P increase, ascribed to enhanced ion bombardment of the surface during growth.

High deposition rates and good quality electrical properties and thickness uniformities over large areas are required for all three films (SiNx, a-Si:H and n+a-Si:H) composing the thin film transistors (TFTs) for the AMLCD industry, while maintaining high tool up-time and low particle formation. Generally these conditions have been achieved for most single-panel multichamber PECVD systems; however, it has become increasingly apparent that a compromise is drawn between the TFT mobility and the deposition rate of the a-Si:H layer. Thus it becomes essential to clearly assess the industry requirements for both deposition rates as well as TFT performance for the different device structures used for AMLCDs, and to discover and control these underlying material properties.

The TEL VHF (40/60 MHz) PECVD system produces high quality, low defect density a- Si:H at deposition rates exceeding 1500 Å/min when analyzed by FTIR, CPM, photo and dark conductivity. Even though the low deposition rate a-Si:H exhibits very similar bulk properties, higher mobility TFTs are produced with a-Si:H deposited at lower RF power. Having both a high ion flux and low ion energy in the SiH4 discharge are likely the most critical conditions for controlling the a-Si:H quality and thus the TFT mobility. Increasing the RF frequency enhances both of these effects, as well as provides a higher deposition rate for a given power density and a higher power threshold for particle/powder formation. For these reasons it is likely a 40/60 MHz plasma will produce better performing TFTs for a given deposition rate when compared with a conventional 13.56 MIHz system. Other process conditions such as diluting the SiH4 in H2 or Ar also seem to play an important role in the optoelectronic properties of the a-Si:H film and ultimately the TFT performance.

The surface morphology of 40-MHz PECVD SiNx films is investigated. We report on the correlation between the deposition conditions, bulk properties, and surface roughness of these TFT insulators. The roughness is measured with atomic-force microscopy, AFM. A link will be presented between the AFM properties and the effects of hydrogen dilution during deposition: gas composition and rf-power-density (P) dependences will be discussed. An increase of the surface roughness to 3.7 nm is observed upon H2 dilution and P increase, ascribed to enhanced ion bombardment of the surface during growth.

High deposition rates and good quality electrical properties and thickness uniformities over large areas are required for all three films (SiNx, a-Si:H and n+ a-Si:H) composing the thin film transistors (TFTs) for the AMLCD industry, while maintaining high tool up-time and low particle formation. Generally these conditions have been achieved for most single-panel multichamber PECVD systems; however, it has become increasingly apparent that a compromise is drawn between the TFT mobility and the deposition rate of the a-Si:H layer. Thus it becomes essential to clearly assess the industry requirements for both deposition rates as well as TFT performance for the different device structures used for AMLCDs, and to discover and control these underlying material properties.

The TEL VHF (40/60 MHz) PECVD system produces high quality, low defect density a- Si:H at deposition rates exceeding 1500 Å/min when analyzed by FTIR, CPM, photo and dark conductivity. Even though the low deposition rate a-Si:H exhibits very similar bulk properties, higher mobility TFTs are produced with a-Si:H deposited at lower RF power. Having both a high ion flux and low ion energy in the SiH4 discharge are likely the most critical conditions for controlling the a-Si:H quality and thus the TFT mobility. Increasing the RF frequency enhances both of these effects, as well as provides a higher deposition rate for a given power density and a higher power threshold for particle/powder formation. For these reasons it is likely a 40/60 MHz plasma will produce better performing TFTs for a given deposition rate when compared with a conventional 13.56 MHz system. Other process conditions such as diluting the SiH4 in H2 or Ar also seem to play an important role in the optoelectronic properties of the a-Si:H film and ultimately the TFT performance.

The demand for lower cost per panel in TFT-LCD production is driving the PECVD market to deposition systems that combine high throughput and uptime with high yield. Today it is generally believed that a multichamber system that combines a number of single-panel deposition chambers is the best way to achieve these goals.

For these PECVD systems to be economical, the deposition rate of a-Si:H, SiNx and SiO2 has to be in the 1200–1500 Å/min range. In 13.56 MHz parallel-plate glow discharge systems SiNx and SiO2 deposition rates exceeding 1500 Å/min are commonly achieved, whereas the deposition rate of a-Si:H is limited to 100–200 Å/min due to powder formation. Over the last 5 years significant progress has been made to increase the deposition rate of a-Si:H. Methods include the use of very-high-frequency glow discharge (VHF-GD) and pulsing of the rf discharge. However, substrate sizes never exceeded 100mm×100mm.

We have developed a multichamber PECVD system for TFT-LCD production where VHFGD is used to obtain uniform high deposition rates for (doped) semiconductors and insulators, such as a-Si:H, n+ a-Si:H, SiNx and SiO2 over areas as large as 470mm×370mm. Even at deposition rates well above 1200 Å/min hydrogen in a-Si:H is exclusively bound as monohydride. The optoelectronic properties of the films are at least as good as those of their 13.56 MHz counterparts and thus good-quality TFTs can be obtained. At the same time the number of added particles is low allowing for high production yields.

Inverted-staggered hydrogenated amorphous-silicon thin-film transistors (a-Si:H TFTs) were deposited in a glow discharge with an excitation frequency of 60 MHz. At 13.56 MHz it has been reported that the field-effect mobility of this type of TFT decreases with increasing deposition rate of the a-Si:H layer, due to an increase in the defect density in the channel. A successful way of increasing the deposition rate without deteriorating the material properties has turned out to be utilizing a higher excitation frequency than the conventional 13.56 MHz.

The deposition rate of the 60-MHz-deposited transistor channel was changed from 350 to 1300 Å/min by diluting the process gas silane with hydrogen and by changing the rf power. The dependence of the a-Si:H material properties on deposition parameters is described. The deposition rate dependence of the mobility in the 60-MHz deposited thin films and devices is presented and discussed in terms of hydrogen dilution in the plasma and the hydrogen content of the a-Si:H films.

The effect of C impurities in a-Si:H in levels of 0.4 to 2.6 at. % were studied over a wide range of metastable defect densities. Three complimentary experimental techniques [electron spin resonance (ESR), drive-level capacitance (DLC) and photo-current] were employed to track the material's defect density with light soaking and annealing, as well as Urbach energies, midgap absorption and mobility gaps energies as a function of the C content. Our results show C impurities have a definite effect on the initial and saturated defect densities, as well as the midgap absorption and Urbach energies at levels 1 at. % and above. The results indicate that C acts mainly as a center for increased disorder in the material which results in an increase in the bandtail widths, and consequently an increase in intrinsic defects. Comparison to the ESR and drive-level data show an excellent agreement between these two techniques in determining the bulk defect densities in a-Si:H.

Bandtail defect carrier trapping dynamics in high quality photo-CVD grown a-Si1-xGex:H (x=0.3 and 0.62) and PECVD deposited a-Si:H films were studied using photo-modulated electron spin resonance (PMESR). Comparison of the dark spin densities with junction capacitance (drive-level profiling) methods indicate that a majority of the defects are neutral in the a-Si:H and 30% Ge samples, but the majority of the defects in the 62% Ge sample are charged. Lineshape deconvolution of the PMESR signals in the in-phase and quadrature phase as a function of photo-modulation frequency, excitation energy and excitation intensity for both intrinsic (state A) and light soaked (state B) states revealed three significant insights: in state A all samples behave similarly, state B behavior of the a-Si:H is dramatically different from the state B alloyed samples, and strong evidence for 'safe hole traps' (SHT) is observed in the state B a-Si:H material.

Comparative studies of low-temperature (T ∼ 30 K) optically induced ESR using above gap red (633 nm) and below gap i.r. (1060 nm) light have been performed on several samples of a-Si:H. Experiments were performed on both films on quartz substrates and on powdered samples which were removed from the substrates. Excitation with red light always yields the well-known light induced ESR (LESR) spectrum which is ascribed to a superposition of two resonances: electrons trapped in conduction band-tail states and holes trapped in valence band tail states in a 1:1 ratio. On the other hand, excitation with i.r. light results in a different LESR spectrum with an enhanced feature overlapping with the signal attributed to trapped band-tail electrons. Although observed on all films studied on quartz substrates, this asymmetry is not observed in a powdered sample where only the band tail electron and hole resonances are observed in the ratio 1:1. This result suggests that interface states are responsible for the excess LESR spin density in the films on quartz substrates. Experiments performed as a function of film thickness confirm this interpretation.

Microwave absorption measurements from 20 to 80 K in magnetic fields up to 12 kG are reported. Below a certain characteristic temperature T* = 80 ± 2 K < Tc the absorption in magnetic-field-cooled samples is smaller and broader in comparison to the zero-field-cooled samples. The incident microwave radiation induces a dc voltage across the sample which is also magnetic field dependent and peaks at zero magnetic field.

We report measurements of an apparent magnetic-field-dependent absorption (imaginary part of the a.c. magnetic susceptibility) in superconducting Y1Ba2Cu3O7 ceramics and crystals. The absorption, which is observed over a wide range of frequencies but only when the material is below the superconducting transition temperature, is characterized by a narrow (∼ 30 Gauss FWHM at 6 MHz) peak and a wide (> 10 kG) feature, both of which are maximum at zero magnetic field. The absorption strength varies approximately as one over the square root of the frequency. The unusual magnetic-field-dependent peaks in the magnetic susceptibility are inherent in single grains and therefore do not originate from intergrain Josephson currents or multigrain (i.e., percolative) loops. The susceptibility peaks must be due to bulk behavior, interactons at grain surfaces, intragrain current loops, or intra-grain Josephson Junctions.