Etching a substrate: processes – Gas phase etching of substrate – With measuring – testing – or inspecting
Reexamination Certificate
2001-07-13
2003-07-01
Powell, William A. (Department: 1765)
Etching a substrate: processes
Gas phase etching of substrate
With measuring, testing, or inspecting
C156S345420, C216S067000, C438S009000
Reexamination Certificate
active
06585908
ABSTRACT:
TECHNICAL FIELD
The present invention relates to semiconductor processing and, more particularly, relates to an apparatus and method for measuring a real-time etching rate.
BACKGROUND
In the fabrication of integrated circuits, the semiconductor substrate or wafer is exposed to numerous process steps. One of the process steps includes etching the materials built up on the wafer to selectively remove certain portions to form the various features utilized in the fabrication of the integrated circuit. The portion removed is defined by a pattern generally formed using an organic photoresist mask. One type of etching process employs a dry chemistry that generally refers to the use of plasma having active species that react with the material to be etched, in order to volatilize and selectively remove the exposed portions. Another type of etching process (sometimes referred to as ashing) reacts with the photoresist to volatilize and strip the photoresist mask from the wafer.
A problem associated with plasma etching (ashing) processes is the difficulty in determining when the etch step has been completed. This difficulty occurs because plasma techniques are typically timed processes, based on predetermined etch rates. The predetermined etch rates are identified by performing a calibration step in which a relative etch rate is determined based on the amount of substrate removed during a known time interval. Since the exact conditions (i.e., pressure, gas flow, electric field, etc.) used during the calibration step may vary to some extent for the etch step during actual device fabrication, timed processes are inaccurate and only provide an estimate as to when the plasma etch process is completed. The time-based process does not provide a real-time etching rate.
As a consequence of the uncertainty in the time needed to etch a wafer, over-etching is used. This usually is defined as a fixed amount of time after which the etch is thought to be complete, in order to guarantee that the etching is complete over the entire wafer. Moreover, time-based processes typically require the use of dedicated equipment for thickness measurements, e.g., an ellipsometer. In order to determine a relative etching rate, before and after thickness measurement must be made thereby requiring operator intervention. More problematic is that the process introduces wafer-to-wafer variability since it is not a real time measurement of the etching rate.
In order to avoid the use of time-based processes for determining the endpoint of an etch step, diagnostic techniques have been developed which analyze the processes occurring in the reaction chamber. One such technique, called optical emission spectroscopy, monitors the intensity of the optical emission from both the plasma and the reactions on the wafer surface. The intensity of the optical emission is related to the concentration of molecular species generated. The completion of the etch process is determined when a change in the intensity of the optical emission is observed. A change in the intensity of the optical emission is observed when the concentration of particular molecular species being monitored is no longer present (or decreases dramatically) indicating that the layer responsible for reacting with the plasma to generate the particular monitored species has been removed. For example, an optical emission signal from hydroxyl species created during etching/ashing of a photoresist layer may be monitored to determine when the photoresist layer has been removed. Optical emission techniques require the reaction chamber to be equipped with an optical port for monitoring the optical emission.
One disadvantage of end-point systems is that the instantaneous ash rate is not known. Also, uncertainty in the end-point determination also requires over-etching.
Another disadvantage of present systems is that one cannot predict end-point. Some applications require the etch process to stop just short of end-point. This is particularly important for manufacturing of thin gate oxides. Unless one knows the instantaneous real-time etch rate, one cannot stop prior to completion with just an end-point system.
Optical interference is another known technique for etching rate measurement. A substrate or wafer with layers of thin material is illuminated with light of a known spectrum. Reflected light from the surface and material interfaces causes an interference pattern that is captured by an optical detector. The interference pattern behavior is determined by the differences in refractive index, thickness of the material being removed, wavelength and angles of incidence. As the thickness of the substrate changes, so does the interference pattern. This method requires the use of an external light source, usually a monochromatic light source such as a laser, dedicated equipment to collect, process and convert the optical interference pattern to a thickness measurement, and dedicated viewing ports in the reaction chamber for both incident and reflected/refracted light. However, the inclusion of such a system may not be a cost effective solution and as is most often the case, viewing ports cannot be arbitrarily located in the reaction chamber as this could impact on critical chamber geometries. For example, incident light and collection at an angle normal to the plane of the wafer requires the viewing port to be located in the same place as the source for the plasma/gas. Moreover, in process chambers using radiant heating to maintain the wafer at elevated process temperatures, the incident light for the optical interference diagnostic needs to be of considerable power so that reflected light is well above the strong background level emitted by the radiant heating source. However, use of such a laser can cause the substrate surface to locally overheat so that the local reaction rate deviates from the wafer average by a non-negligible amount. In this sense, the technique can no longer be considered non-perturbative.
Referring now to
FIG. 1
, there is shown a figure illustrating the general principles of optical interference for thin film coatings on reflective substrate materials. A semiconductor wafer
10
coated with a photoresist layer
12
having a thickness d and a refractive index n.
When an external light beam &lgr; is projected over the photoresist surface, light is both reflected and refracted from the surface. The reflected light beam (
1
) and refracted light beams (
2
,
3
,
4
, . . . ) travel different distances depending on the refractive index of the material comprising the surface and the thickness of the photoresist layer. Assuming the thickness d to be constant throughout the length of relevant refractions, the difference in the distance traveled by consecutive refracted beams (
2
,
3
,
4
. . . ) is L. This relationship can be described mathematically as shown in equations (1) and (2). For a monochromatic light source &lgr;, the distance L corresponds to a phase shift &Dgr;&phgr; between consecutive beams in accordance with well known optical principles. It should be noted that what really matters is the difference in “optical” path length and not just geometric path length, since the light ray travels more slowly in materials of higher index of refraction. The optical path length depends on both the geometric path length and the index of refraction along that path.
L
=2d{square root over (
n
2
−sin
2
)}&agr; (1)
&Dgr;&phgr;=2&pgr;(
L
/&lgr;) (2)
If the beams
1
and
2
are in phase with one another at the detector, the beams produce a constructive interference pattern, i.e., &phgr;=2k&pgr; (k integer). Conversely, if the beams are out of phase, the beams will provide a destructive interference pattern, i.e., &Dgr;&phgr;=(2k+1)&pgr;. That is, a minimum for one of the beams coincides with a maximum for the other beam, or vice versa, thereby canceling or subtracting each other out.
When the beams are projected onto a target, e.g., a photo detector or optical fiber, the phase shift &Dgr;&phgr; will cause th
Cardoso Andre G.
Janos Alan C.
Axcelis Technologies Inc.
Cantor & Colburn LLP
Powell William A.
LandOfFree
Shallow angle interference process and apparatus for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Shallow angle interference process and apparatus for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Shallow angle interference process and apparatus for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3046315