Thin film CdS has been widely used in thin-film photovoltaic devices. The most efficient Cu(In, Ga)Se2 (CIGS) solar cells reported to date utilized a thin CdS buffer layer prepared by a reactive solution growth techni...
Thin film CdS has been widely used in thin-film photovoltaic devices. The most efficient Cu(In, Ga)Se2 (CIGS) solar cells reported to date utilized a thin CdS buffer layer prepared by a reactive solution growth technique known as chemical bath deposition (CBD). Considerable effort has been directed to better understand the role and find a replacement for the CBD CdS process in CIGS-based solar cells. We reported a low temperature (∼150 °C) Metalorganic Chemical Vapor Deposition (MOCVD) CdS thin film buffer layer process for CIGS absorbers. Many prior studies have reported that CBD CdS contains a mixture of crystal structures. Recent investigations of CBD CdS thin films by ellipsometry suggested a multilayer structure. In this study we compare CdS thin films prepared by CBD and MOCVD and the effects of annealing. TED and XRD are used to characterize the crystal structure, the film microstructure is studied by HRTEM, and the optical properties are studied by Raman and spectrophotometry. All of these characterization techniques reveal superior crystalline film quality for CdS films grown by MOCVD compared to those grown by CBD. Dual Beam Optical Modulation (DBOM) studies showed that the MOCVD and CBD CdS buffer layer processes have nearly the same effect on CIGS absorbers when combined with a cadmium partial electrolyte aqueous dip.
Ultrathin devices have become important because they can be flexible and can be included on a plastic card. The low thickness increases the amount of surface scattering and scattering on dislocations and on other latt...
Ultrathin devices have become important because they can be flexible and can be included on a plastic card. The low thickness increases the amount of surface scattering and scattering on dislocations and on other lattice defects caused by bending. This goes hand in hand with a reduction of lattice scattering, both normal and umklapp or intervalley. For ultrasmall devices, the conventional quantum 1/f effect usually determines the low-frequency noise. This effect is proportional to the squared velocity change of the current carriers which is large in lattice scattering, and particularly large in umklapp or intervalley scattering. Therefore, we expect a reduction of 1/f noise with decreasing thickness of the devices. Nevertheless, there often will be an increase in noise, due to the general 1/N dependence on the number of carriers N which define the electric current and the noise current, and due to the additional noise caused by current inhomogeneities introduced by bending-related defects. Finally, as we show in this paper, the 1/N dependence is often compensated by the presence of a coherent quantum 1/f effect vestige.
The use of a physically-based electromigration model as a means of validating design rules in aluminum-based microelectronic interconnects is presented. This model consider the migration of atoms through the line when...
The use of a physically-based electromigration model as a means of validating design rules in aluminum-based microelectronic interconnects is presented. This model consider the migration of atoms through the line when driven by an applied current and mechanical stress gradients. The model is then applied to a specific interconnect geometry common to CMOS logic circuits. Here the saturation resistance in the interconnect is shown to agree well with experimental observations. The value of the effective charge (Z*) in this material is found to be −8.8. These results demonstrate the usefulness of electromigration modeling as a means of developing interconnect design rules.
The Monte Carlo ion implant simulator UT-MARLOWE has usually been verified using a large array of Secondary Ion Mass Spectroscopy (SIMS) data (∼200 profiles per ion species)(1). A model has recently been developed (1...
The Monte Carlo ion implant simulator UT-MARLOWE has usually been verified using a large array of Secondary Ion Mass Spectroscopy (SIMS) data (∼200 profiles per ion species)(1). A model has recently been developed (1) to explicitly simulate defect production, diffusion, and their interactions during the picosecond ‘defect production stage’ of ion implantation. In order to thoroughly validate this model, both SIMS and various damage measurements were obtained (primarily channeling-Rutherford Backscattering Spectroscopy, Differential Reflectometry and Tapered Groove Profilometry, but supported with SEM and XTEM data). In general, the data from the various experimental techniques was consistent, and the Kinetic Accumulation Damage Model (KADM) was developed and validated using this data. This paper discusses the gathering of damage data in conjunction with SIMS in support of the development of an ion implantation simulator.
Electromigration failure under DC stress has been studied for more than 30 years, and the methodologies for accelerated DC testing and design rules have been well established in the IC industry. However, the electromi...
Electromigration failure under DC stress has been studied for more than 30 years, and the methodologies for accelerated DC testing and design rules have been well established in the IC industry. However, the electromigration behavior and design rules under time-varying current stress are still unclear. In CMOS circuits, as many interconnects carry pulsed-DC (local VCC and VSS lines) and bidirectional AC current (clock and signal lines), it is essential to assess the reliability of metallization systems under these conditions. Failure mechanisms of different metallization systems (Al-Si, Al-Cu, Cu, TiN/Al-alloy/TiN, etc.) and different metallization structures (via, plug and interconnect) under AC current stress in a wide frequency range (from mHz to 500 MHz) has been study in this paper. Based on these experimental results, a damage healing model is developed, and electromigration design rules are proposed. It shows that in the circuit operating frequency range, the “design-rule current” is the time-average current. The pure AC component of the current only contributes to self-heating, while the average (DC component) current contributes to electromigration. To ensure longer thermal-migration lifetime under high frequency AC stress, an additional design rule is proposed to limit the temperature rise due to self-joule heating.
The presence of charged dust in a plasma can lead to very low frequency dust acoustic waves and instabilities. In certain laboratory plasmas the dust is strongly coupled, as characterized by the condition Γd=Qd2 exp(...
The presence of charged dust in a plasma can lead to very low frequency dust acoustic waves and instabilities. In certain laboratory plasmas the dust is strongly coupled, as characterized by the condition Γd=Qd2 exp(−d/λD)/dTd⩾1, where Qd is the dust charge, d is the intergrain spacing, Td is the dust thermal energy, and λD is the plasma screening length. When the dust is strongly coupled, the spatial correlation of the grains can affect the dispersion relation of these waves. We review our recent work [1] on the dispersion properties of dust acoustic waves in the strongly coupled (liquid) phase in a dusty plasma, including also the effects of dust-neutral collisions. We then discuss a preliminary analysis of the effect of strong dust coupling on an ion dust two-stream instability in a collisional dusty plasma. Applications to laboratory dusty plasmas are discussed.
Whereas it is generally acknowledged that code tangling reduces the quality of software and that aspect-oriented programming (AOP) is a means of addressing this problem, there is — as yet — no clear definition or ch...
详细信息
In this paper, a survey of a particular class of unsupervised learning rules for neural networks is presented. These learning rules are based on variants of Hebbian correlation learning to update the connection weight...
In this paper, a survey of a particular class of unsupervised learning rules for neural networks is presented. These learning rules are based on variants of Hebbian correlation learning to update the connection weights of two-layer network architectures consisting of an input layer with n units and an output layer with m units. It will be demonstrated that the networks are able to perform a variety of important data analysis tasks, including Principal Component Analysis (PCA), Minor Component Analysis (MCA) and Independent Component Analysis (ICA).
Ultra-thin (0.4∼0.8 nm) silicon nitride films were deposited by RPECVD (Remote Plasma Enhanced Chemical Vapor Deposition) onto thin oxides for the p+-polysilicon gated PMOS devices. Boron penetration is suppressed wh...
Ultra-thin (0.4∼0.8 nm) silicon nitride films were deposited by RPECVD (Remote Plasma Enhanced Chemical Vapor Deposition) onto thin oxides for the p+-polysilicon gated PMOS devices. Boron penetration is suppressed when the top nitride layer thickness increased to at least 0.8 nm. Suppression of boron diffusion is monitored by flat band voltage shifts as well as polysilicon depletion. Compared to oxides, improved charge to breakdown (Qbd) is obtained for nitride/oxide dual layers, while interface state density remains at a level of 3×1010 eV−1 cm−2.
A phase-shifting point diffraction interferometer (PS/PDI) has recently been developed to evaluate optics for extreme ultraviolet (EUV) projection lithography systems. The interferometer has been implemented at the Ad...
A phase-shifting point diffraction interferometer (PS/PDI) has recently been developed to evaluate optics for extreme ultraviolet (EUV) projection lithography systems. The interferometer has been implemented at the Advanced Light Source at Lawrence Berkeley National Laboratory and is currently being used to test experimental EUV Schwarzschild objectives. Recent PS/PDI measurements indicate these experimental objectives to have wavefront errors on the order of 0.1 waves (∼1 nm at a wavelength of 13.4 nm) rms. These at-wavelength measurements have also revealed the multilayer phase effects, demonstrating the sensitivity and importance of EUV characterization. The measurement precision of the PS/PDI has been experimentally determined to be better than 0.01 waves. Furthermore, a systematic-error-limited absolute measurement accuracy of 0.004 waves has been demonstrated.
暂无评论