Computer-aided design and analysis of circuits and semiconductor – Nanotechnology related integrated circuit design
Reexamination Certificate
2000-02-02
2002-09-24
Smith, Matthew (Department: 2825)
Computer-aided design and analysis of circuits and semiconductor
Nanotechnology related integrated circuit design
C716S030000
Reexamination Certificate
active
06457169
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to methods for measuring the misalignment, or overlay error, between the patterns created on a wafer in different steps of the integrated circuit manufacturing process, and more specifically to improved methods of measuring misalignment which utilize geometric phase analysis.
2. Description of the Related Art
During the process of integrated circuit fabrication, several hundred processing steps may be required as thin layers of semiconductors, insulators and metals are deposited and patterned. The patterns in each layer of the integrated circuit are created using the process of lithography. A photosensitive resist (“photoresist”) is deposited on the silicon wafer and exposed to light through a mask carrying the desired pattern. The photoresist is then developed and removed from areas corresponding to the pattern, allowing the pattern to be transferred by etching into the layer below.
With every generation of smaller and faster integrated circuits, the size of features in the masks is reduced, and the accuracy with which features in each layer must be aligned with those formed in previous layers must increase. Therefore, the problem of measuring the misalignment between features formed in different layers has become more critical. State of the art 64 Mb dynamic random access memory (DRAM) chips contain circuits with features as small as 250 nanometers (nm), and performance of the circuit is affected if one set of features is misaligned by more than about 10 nm from the set of features formed in previous processing steps. The placement precision, or cumulative difference between patterns from various mask levels, is commonly called “overlay error”.
To measure the amount of misalignment, or overlay error, the industry currently makes use of an alignment technique illustrated in
FIGS. 1A and 1B
. Each mask includes several alignment areas, separate from the wiring or other pattern of the circuit. These alignment areas contain alignment marks which are cross-shaped or chevron-shaped features, usually several microns in length.
FIG. 1A
shows a plan view of such an alignment area.
The first alignment mark is etched onto the wafer at the first mask level (referred to hereafter as level “A”) and subsequent levels are aligned with respect to this mark. In the example illustrated here, the alignment mark for level “A” consists of four large chevrons (labeled “
10
” in
FIG. 1A
) arranged at the corners of a 50×50 micron square. After level “A” is patterned using the mask, the chevrons are easily visible in an optical microscope. The next mask (referred to as level “B”) is then exposed. Level “B” has a smaller set of chevrons (labeled
11
) which, if accurately aligned, will fit within the larger chevrons of level “A”. After exposure and development of the photoresist for level “B”, the alignment mark for level “B” is faintly visible in the photoresist. By measuring the positions of the edges of the chevrons along a line L-L′, as shown in
FIG. 1B
, the degree of misalignment (labeled
12
in
FIG. 1B
) in the x direction can be measured. A similar measurement is then done in the y direction. If the error is beyond acceptable tolerances, the photoresist is washed off and the exposure of the level B mask is repeated.
This technique was originally developed for much larger feature sizes, but after several years of optimization this technique is now being used to measure overlay error with an accuracy of 10 nm. This accuracy is sufficient for present day needs but will be inadequate for future generations of integrated circuits with smaller feature sizes requiring more precise alignment. For these smaller circuits, yield problems occur when the conventional technique measures an acceptable overlay error and allows the wafer to be fully processed, but it is later found that there is poor performance due to an overlay error smaller than can be resolved using the technique.
A separate problem of the conventional technique is that it relies on measurements of large alignment marks, several microns in size, whereas the features in the actual circuits are much smaller, of the order of 250 nm (or 0.25 microns). Large alignment marks are used because they are visible in an optical microscope and therefore can easily be measured: the optical microscope can resolve features down to about 250 nm. However, since the processing is optimized for the smaller features in the circuit, the edges of the large alignment marks may be improperly etched, making it difficult to measure their positions. More importantly, it may be difficult to relate the position of the large alignment marks to the positions of the small features in the actual circuits, because large and small features are etched at different rates during processing.
To improve measurement accuracy, and to avoid the problems associated with using large alignment marks, several other overlay measurement techniques have been suggested, such as techniques based on moire fringes. In moire fringe techniques, the alignment marks are large areas filled with a regular pattern of small lines or dots.
In a typical example of a moire fringe technique, an alignment mark consisting of a regular pattern of dots is etched onto the wafer at the first mask level (again referred to as level “A”). The next mask level “B” is then exposed. The alignment mark for level “B” also consists of a regular pattern of dots and these are superimposed over the level “A” pattern. However, the level “B” pattern has a slightly different spacing between dots. Because of this difference in spacing, in some places the “A” and “B” dots are in registry, i.e. superimposed exactly, but in other places they are out of registry. Although individual dots are too small to see clearly in an optical microscope, the areas where dots are in and out of registry (the “moiré fringes”) can be distinguished optically. Visual inspection of the position of these areas is used to determine the overlay error.
By using alignment marks which are made up of many small dots or lines, moiré fringe techniques avoid the problems associated with using large features for alignment marks. However, other significant problems occur with moiré fringe techniques. Three problems are worth noting. Firstly, once the level “A” pattern of dots is etched onto the wafer, it forms a non-planar surface on which the level “B” pattern must be placed. This can distort the level “B” pattern and make it difficult to expose and develop correctly. A second problem is that the level “B” pattern shows very weak contrast (since it is only patterned in photoresist) compared to level “A” (which is etched onto the wafer). When superimposed onto the high contrast level “A” pattern, level “B” is very difficult to see, making the moiré fringes weak and the measurement less precise. A third problem is that the difference in spacing between levels “A” and “B” must be small (around 1%) and carefully controlled in order to get appropriate moiré fringes and this can be difficult to achieve.
The invention to be described below is significantly different from both the moiré and conventional techniques which have just been described, in that it uses a different type of alignment mark combined with mathematical analysis to measure the overlay error. This enables the problems described above to be avoided while also allowing an improved accuracy in the measurement of overlay error.
SUMMARY OF THE INVENTION
The object of the present invention is to provide a structure and method for measuring the overlay error between at least two mask levels, each of which includes specific alignment marks. These alignment marks consist of repeating patterns of small dots or other features. All marks have the same spacing between features, and alignment marks from successive mask levels are placed next to each other. Mathematical analysis of the aligmnent marks is then used to calculate the overlay error with a high degree of accuracy. For this, the invention make
August Casey P.
Garbowski Leigh Marie
International Business Machines - Corporation
McGinn & Gibb PLLC
Smith Matthew
LandOfFree
Geometric phase analysis for overlay measurement does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Geometric phase analysis for overlay measurement, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Geometric phase analysis for overlay measurement will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2914089