Patent for License:System for Process Variation Monitor
SUMMARY OF THE TECHNOLOGY
The technology is, in essence, of a method to extend the process monitoring capabilities of a semiconductor wafer optical inspection system so as to be able to detect, and possibly quantify, macro-defects, i.e. low-resolution effects of process variations over the surface of a wafer, at much higher sensitivity than possible by analyzing micro-defects conventionally detected in such a system; optionally it enables also detecting effects of temporal process variations on successively processed wafers. The method is preferably operative simultaneously, and in conjunction, with micro-defects detection operation of such an optical inspection system and, in common with it, involves examining the entire inspectable surface of the wafer. The method of the technology is designed to detect process variations of lower magnitude than such that would result in a significant number of detectable micro-defects. It is noted that the simultaneity feature and the feature of entire surface examination, mentioned above, advantageously contrast with the mode of operation of process monitoring methods of current art, whereby geometric effects of process variations are examined only at those areas in which micro-defects have been detected and, moreover, such examination is carried out at a so-called review phase, which is separate from the defects detection operation and during which a higher-resolution scan is required--possibly even using different equipment.
In one embodiment of the invented system, partial results of the defects detection operation are further processed to directly obtain indications of process variations over the entire inspected area. Also in common with the defects detection operation, the invented method can advantageously use the outputs of any multiple sensors comprised in the inspection system, such as those disposed along various angles to the inspected surface and to the illuminating beam, including sensing a so-called dark field or a so-called bright field, to sense more varied effects and thus obtain more accurate or more reliable results. The invented method can also form a basis for a capability of classifying process variations whose effects have been detected and measured by it, although the embodiments to be described below do not include such a capability.
Optionally and with obvious modifications, the invented method can function separately from, or independently of, defects detection operation, although, it is noted, combined operation is generally more economical. The technology is also of equipment and system operative to carry out the method as disclosed herein.
It will be appreciated that, though the present disclosure describes the technology in terms of inspecting semiconductor wafers, being processed to become integrated-circuit dies, the technology is equally applicable to the inspection of surfaces of other substrates, such as those carrying photonic devices or those undergoing any other processing, as well as surface variations not necessarily ascribable to a process. It will be appreciated that the technology is equally applicable to the inspection of surfaces by means other than optical, such as an electron- or ion beam, and, in general, to any inspection system whereby the surface is probed or sensed point-by-point. In any such system each sensor outputs intensity values that correspond to energy received by the sensor as a result of reflection of the probing beam from the inspected surface. These values are generally called radiation intensity values, but in the sequel will also be referred to, interchangeably, as light intensity values (since the preferred embodiments utilize a light beam for the probing).
The invented method, as applicable to a wafers inspection system, basically comprises the following logical steps: (a) Obtaining one or more light intensity values for each point (pixel) on the inspected surface, belonging to corresponding classes, such as the outputs of the various sensors in the inspection system; these may be identical to the values used for micro-defects detection; (b) preferably calculating for each pixel one or more derived values; (c) dividing the surface into an array of geometric blocks, each block including a considerable plurality of contiguous pixels; (d) calculating for each block, as a whole, a so-called signature, which is a set or an array of variables, as a function of the light intensity values and the derived values of its several pixels; (e) for each block, comparing its signature with a designated comparison signature, possibly associated with a comparison block, and thereby calculating one or more process deviation indications.
The derived values in step b are preferably local spread values, i.e. a measure of the extent of variability of intensity values in the immediate vicinity of the referenced pixel; they are calculated separately with respect to each class (i.e. sensor).
In one embodiment of the method, calculating a signature includes calculating for each pixel, and, a with respect to each class, a histogram over pairs of light intensity- and spread values. In one configuration, there is provided for each block over the surface of the wafer a model comparison signature and the comparing in step e is with respect to that model signature. In another configuration, the comparison with respect to any block, is between any signature calculated for the currently inspected wafer and a corresponding one calculated for the previously inspected one. In yet another configuration, applicable to the prevalent case that the IC being fabricated is an array of identical dies, the array of blocks is defined in alignment with the array of dies, there being a plurality of blocks over each die; the comparing of signatures is between any block and a congruent block over one or more other dies.
In another embodiment of the invented method, applicable particularly when also micro-defects are being detected, there are first obtained for each pixel, and corresponding to each sensor, a light intensity value and possibly a local spread value--both in comparison with corresponding ones for a congruent pixel on another die. The differences yielded by such comparisons are preferably thresholded according to a given threshold curve and each excess is noted; the threshold values of the given curve are typically much lower than those applied in a similar procedure during detection of defects. Finally, the excesses, corresponding to each sensor, are counted over each block, the sums forming the respective process deviation indications.
Class 700: Data Processing: Generic Control Systems Or Specific Applications
This class is structured into two main divisions: (1)for the combination of a data processing or calculating computer apparatus (or corresponding methods for performing data processing or calculating operations) AND a device or apparatus controlled thereby, the entirety hereinafter referred to as a "control system". (2)for data processing or calculating computer apparatus (or corresponding methods for performing data processing or calculating operations) wherein the data processing or calculating computer apparatus is designed for or utilized in a particular art device, system, process, or environment, or is utilized for the solution of a particular problem in a field other than mathematics (arithmetic processing per se is classified elsewhere).Subclass 121: Integrated circuit production or semiconductor fabrication
Class 348: Television
Generating, processing, transmitting or transiently displaying a sequence of images, either locally or remotely, in which the local light variations composing the images may change with time.Subclass 14.01: TWO-WAY VIDEO AND VOICE COMMUNICATION (E.G., VIDEOPHONE)
Class 382: Image Analysis
This is the generic class for apparatus and corresponding methods for the automated analysis of an image or recognition of a pattern. Included herein are systems that transform an image for the purpose of (a) enhancing its visual quality prior to recognition, (b) locating and registering the image relative to a sensor or stored prototype, or reducing the amount of image data by discarding irrelevant data, and (c) measuring significant characteristics of the image.Subclass 145: Inspection of semiconductor device or printed circuit board