Radiant energy – Inspection of solids or liquids by charged particles – Methods
Reexamination Certificate
2002-09-23
2004-08-17
Lee, John R. (Department: 2881)
Radiant energy
Inspection of solids or liquids by charged particles
Methods
C250S442110
Reexamination Certificate
active
06777674
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to techniques for removing and analyzing microscopic particles from a sample surface, particularly from semiconductor samples.
BACKGROUND
In the semiconductor industry, unexpected particles due to contamination will cause yield loss during the manufacturing process. Since a major focus of this industry is aggressive reduction in feature size for pattern line widths, the minimum size of particles that can cause performance loss also decreases rapidly. A reasonable estimate for a “killer defect” size is that greater than one-third the size of the smallest feature on the semiconductor wafer.
Although semiconductor manufacturing is performed in clean rooms with stringent particle standards, unexpected contamination will occur due to sources such as moving parts, human presence, gas condensation, and chamber wear. Control and removal of these particles is a continuous process. In many cases, removal of the source of particles requires an understanding of their origin. Many of these particles or defects are too small for detection in a general purpose optical inspection microscope, so higher resolution methods are required, using charged-particle microscopes, such as scanning electron microscopes (SEM), transmission electron microscopes (TEM), scanning Auger microprobes (SAM), or focused ion beam (FIB) instruments, are required.
Even an image of the particle is usually insufficient to trace the origin of the particle, and more information is required. Elemental composition is valuable in identifying the defect. This can be done in various ways using the charged-particle systems mentioned above. Unfortunately, most of the analytical methods are limited by background signals from the environment of the particle.
Throughput is also a critical parameter in semiconductor manufacturing. Existing strategies for compositional analysis of particles on a semiconductor wafer, for example, usually require removal of the wafer from the fabrication area for off-line analysis using methods such as those described below. Removal from the line severely reduces the throughput of the manufacturing process.
Particle identification on sample surfaces using electron-beam based identification is complicated by the size of the particle relative to the electron penetration depth, and by the nature of surrounding materials in the sample. As an electron beam interacts with bulk solid materials, it expands to fill a teardrop-shaped volume as it loses energy. As the primary beam interacts with atoms in this volume, it generates low energy Auger electrons and X-rays that are characteristic of the elements involved.
The particular X-ray line generated will depend on the atomic number of the element, the energy of the electron during the interaction, and other factors. When trying to identify an unknown particle using conventional Energy-dispersive X-ray Spectrophotometry (EDS), the energy of the electron beam must be large enough to generate inner-shell X-rays from all possible relevant elements, which, for semiconductor applications, may include elements of high atomic number such as tungsten. Unfortunately, this energy results in a penetration depth that may be much larger than the particle of interest, resulting in X-ray generation from the sample surface. These X-rays interfere with any signal from the particle, making unique identification of the particle material difficult. Conventional strategies for solving this problem involve either resolving the X-ray lines of different elements, or reducing the energy of the exciting electron beam.
For example, it is possible to detect and analyze electron-beam generated X-rays from a particle by measuring the intensity and diffraction angle of the X-rays diffracted by a reference crystal, or Wavelength Dispersive X-ray Spectrometry (WDS). One chooses the crystal atomic spacing to deflect (with very high resolution) X-rays of a given energy, thus allowing separation between X-ray lines of different elements. This method has higher energy resolution than EDS but much slower throughput. In addition, if the particle could be, as it often is, of the same composition as the sample surface, this method will not uniquely determine the particle composition.
Other solutions involve reducing the energy of the primary electron beam to guarantee the activated volume is less than the volume of the particle of interest. This reduction in primary electron-beam energy results in characteristic X-rays of much lower energy (M or L shell X-rays, rather than K shell). Conventional cooled semiconductor-based detectors use the generation and collection of electron-hole pairs as a measure of the energy of the ionizing radiation (a few eV for each electron-hole pair, depending on the detecting material). A reduction in the X-ray energy therefore leads to a reduced number of electron-hole pairs and reduced sensitivity to the particle material. In addition the resolution of these detectors is governed by the statistics of the electron-hole generation process, and reducing the energy of the detected X-ray often leads to ambiguous identification of the element of interest. X-ray micro-calorimeter methods have been used to detect these weak X-ray signals, using heat transferred to the detector rather than the generation of electron-hole pairs. This process does allow measurement of small X-ray energies, but micro-calorimeter instruments are expensive, have complicated cooling requirements, and are slow compared to other methods. Also, the electron beam must be kept smaller than the smallest dimension of the particle of interest, rendering the method impractical for small, unsymmetrical particles.
Scanning Auger microprobe analysis also uses an electron beam to irradiate a particle of interest, but rather than detecting any X-rays generated it focuses on the detection of Auger electrons ejected from the atoms of the material. These Auger electrons come from outer shells and have relatively low energies. The Auger electron energies from a material produce a pattern that is characteristic of each element in the material, and the shape and exact energy of the Auger transitions provide information on the chemical bonding of the elements in the material (such as, phase or compound information). The escape depth of these electrons is quite small (a few nm), so Auger analysis focuses mainly on the surface of a sample. This is an advantage for the analysis of small diameter particles (<10 nm). For the analysis of larger particles, one can generate depth profiles by using an ion beam to sputter through the particle and take periodic measurements, but this is inherently destructive of the surrounding sample due to ion milling in the SAM, and requires background analyses on the sample near the location of the particle. Auger analysis is typically more sensitive to light elements than standard EDS analysis, making it more suitable to identify organic materials. However, to improve counting statistics, high electron beam currents are typically employed. This exaggerates the issues of thermo-mechanical drift and drift due to electrical charging of the sample. This means that operating the SAM in the “spot mode,” with the electron beam positioned on the particle, involves a risk that over time the electron beam spot will drift onto the sample that surrounds the particle. And the use of a raster pattern for the electron beam will be more tolerant of drift for keeping the beam on the particle, but will involve significant contamination of the results with signal from the surrounding material. In either case, background contamination of the Auger results is a serious issue, and Auger analyses of the surrounding material are required to uniquely identify the signal from the particle. The acquisition of background analyses reduces throughput and inherently damages the sample.
TEM can often be used for analysis of particles in or on surfaces. There are a variety of methods for isolating the particle for analysis, including replication, lift-out or cross sectioning the a
Anthony John M.
Moore Thomas M.
Lee John R.
Leybourne James J.
Omniprobe, Inc.
Thomas John A.
LandOfFree
Method for manipulating microscopic particles and analyzing does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for manipulating microscopic particles and analyzing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for manipulating microscopic particles and analyzing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3341685