The present invention relates to techniques for finding out defects in a semiconductor device (product or parts) during its manufacturing process and for coping with them.
In the manufacturing process of a semiconductor device such as semiconductor wafer, photomask and magnetic disc, foreign matters on the surface of the semiconductor device and pattern defects as well are responsible for causes of a faulty product and search and clarification of failure root causes of faults is important. To this end, it is necessary to quantify the foreign matter and pattern defect (hereinafter referred to as appearance fault) with a view to constantly monitoring the presence or absence of problems in the manufacturing apparatus and environment. In addition, confirmation as to whether or not the appearance fault has a fatal influence upon a product needs to be made by observing the shape or contour of the appearance fault.
The observation work has hitherto been carried out with the help of human eye in many cases. Accordingly, there arises a problem that the position and kind of a defect is biased depending on an observation attendant person or the defect to be observed is not determined definitely. Lately, for the sake of solving the problematic points as above, the introduction of techniques of automatic defect review (ADR) and automatic defect classification (ADC) has been started. For example, a system has been proposed according to which during observation or review (defect review) of an inspected parts (for example, a pattern formed on a wafer) by using an observation unit of scanning electron microscope (SEM) type, the work can proceed while reducing loads imposed on an operator (see U.S. Pat. No. 6,259,960B1, for instance).
Further, in recent years, the critical dimension in working or fabricating semiconductor devices has been minified and the defect has also been minified correspondingly. Therefore, there urgently arise needs for changing conditions of inspection in an inspection unit adapted to draw out defects and for outputting at a time a plurality of defects extracted under individual inspection conditions. On the other hand, as the sensitivity of the inspection unit increases, the output of the inspection unit becomes highly vulnerable to noise and the number of defective points detected per one process of inspection may sometimes exceed several tens of thousand. Then, for elimination of the noise, a method has been known in which defects now under inspection are sorted pursuant to the real-time defect classification (RDC) function and then the noise is eliminated.
Furthermore, in order to determine inspection conditions in the inspection unit and conditions in using the RDC function for noise elimination, a technique has been proposed according to which as many as possible pieces of information delivered out of the inspection unit, information about identification (ID) and coordinates of a defect delivered out of the observation unit and ADR information and ADC information also delivered out of the observation unit as well are readjusted to facilitate the defect analysis (see JP-A-2001-156141, for instance).
But, by merely reviewing defects found out by the inspection unit as in the aforementioned related arts, causes of defects cannot be grasped accurately in many cases. Accordingly, the attendant person watches the distribution of defects and if suspicious defects in critical dimension, alignment (register of positions of individual layers) and film thickness are questioned, the measured pieces of data are confirmed to search causes of the defects. The attendant person, however, is not always near the apparatus and when the attendant person is unattended, for example, at night, the semiconductor device (such as wafer) is permitted to proceed to the next process and decisive evidences cannot often be acquired on the spot, leading to a delay in taking countermeasures against the defect. Even when confirmation after the fact is desired to be made, the measurement as above is not conducted for all wafers and all dies but is conducted only for sampled wafers and dies and accordingly, the defective portion in want of knowledge cannot be known directly but can merely be estimated from temporally and spatially close data in many cases.
Under the circumstances, the present invention contemplates the elimination of the conventional problems as above and its object is to provide defect review method and apparatus which can acquire information profitable to post-review when a defect is found in, for example, the semiconductor device in the course of manufacture.
To accomplish the above object, a defect review method according to this invention is based on a defect review apparatus having a storage for receiving and storing defect information concerning an inspection objective captured by a wafer inspection system, an image acquisition unit for capturing images concerning the inspection objective and a process unit for acquiring data for defect review based on the defect information by using the image acquisition unit.
The process unit makes a decision as to whether a cluster representative of a set or gang of defects exists in the defect information read out of the storage, and when the presence of the cluster is determined, acquires an image of a defective portion forming part of the cluster and additional data in respect of the inspection objective by using the image acquisition unit on the basis of a distributive feature of the cluster.
Other expedients will be described later.
According to the present invention, when a defect in the semiconductor device or the like is found out in the manufacturing process, information profitable for post-review can be obtained.
Other objects, features and advantages of the invention will become apparent from the following description of the embodiments of the invention taken in conjunction with the accompanying drawings.
Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. In the present embodiment, a review apparatus (defect review apparatus) of the present invention will be described as being applied to the production line of semiconductor devices and a defect review method will be described as being practiced on the semiconductor device production line.
Referring first to
The wafer inspection system 1 can be materialized with, for example, a brightfield wafer inspection tool, a darkfield wafer inspection tool, an SEM type wafer inspection tool or a charge coupled device (CCD) camera. Details of the review apparatus 2 will be described later. The wafer inspection system 1 and review apparatus 2 are illustrated by dashed line in
The wafer inspection system 1 and review apparatus 2 are connected through a communication line 4 to a data process unit 3 representing a destination to which inspection data and image data are transmitted. Wafers for a product flow in a unit of lot along the manufacturing processes 11. A wafer having gone through all of the manufacturing processes 11 undergoes a probe test in a probe inspection unit 12. After having been processed through the manufacturing processes 11 in connection with which the execution of the appearance inspection is prescribed in advance, the wafer is conveyed to a place where the wafer inspection system 1 is located by a work person or by means of a conveyer.
Turning to
Defect information 21 (appearance data) obtained through the appearance inspection by the wafer inspection system 1 is transmitted to the data process unit 3 and is stored, together with a lot number, a wafer ID, an inspection process and an inspection date, in the storage 32 of data process unit 3. An example of the defect information 21 is illustrated in
Reverting to
Since pieces of the defect information 21 delivered out of the wafer inspection system 1 are enormous, defect information 22b or 23b drawn out by the data process unit 3 through a plurality of filter functions is sent to the optical review unit 24 or SEM type review unit 25 via the communication line 4. The format of defect information 22b or 23b may be the same as or different from that of defect information 21.
On the basis of the extracted defect information 22b or 23b, an image of the defect portion is captured in the optical review unit 24 or SEM type review unit 25 and by using the image, defect classification is carried out with ADC function loaded in each tool constituting the review apparatus 2. The thus obtained information is transmitted as ADR/ADC information 22a or 23a to the data process unit 3 by way of the communication line 4 and stored in the storage 32. The ADR/ADC information 22a or 23a is obtained when the defect in the semiconductor device is found out in the manufacturing process 11. This information is useful for post-review and the user can watch it on the display 33. The process for obtaining the information useful for review will be explained below by using first and second process examples.
By making reference to
Firstly, in step S100, the review apparatus 2 captures appearance inspection data from the data process unit 3.
Next, in step S101, the review apparatus 2 conducts a process of cluster recognition on the basis of defect information 21 in the appearance inspection data. The cluster recognition process can be practiced through a method shown in, for example, the JP-A-2005-197629. The cluster referred to herein signifies that a plurality of defects form a set or gang in a specified relationship and it means a gang of the defects (also termed a cluster defect).
In step 102, the review apparatus 2 makes a decision as to whether a cluster exists. If the absence of any cluster is determined (No in step S102), the review apparatus ends the process.
If the presence of clusters is determined (Yes in step S102), the review apparatus 2 makes a decision as to whether dependency (pattern-wise or trend-wise) prevails in respect of the individual clusters within the wafer surface in step S103 as will be detailed later with reference to
If the presence of the dependency prevailing within the wafer surface is determined (Yes in step S103), the review apparatus 2 captures images at the same sites in a defective die and a normal die (mutually corresponding sites), respectively, in step S104 as will be detailed later with reference to
Following No in the step 103 or following the step S104, the review apparatus 2 decides whether dependency prevails within a die in respect of the individual clusters in step S105 as will be detailed later with reference to
If the presence of the dependency within the die is determined (Yes in step S105), the review apparatus 2 captures images at a defective site (defective portion) and a normal site, respectively, in the objective die in step S106 as will be detailed later with reference to
Following No in the step S105 or following the step S106, the review apparatus 2 decides whether the percentage of foreign matters (the ratio between the numbers of dies in which foreign matters exist) exceeds A % (predetermined value) in respect of the wafer having individual clusters (in case the review apparatus 2 lacks the ADC function, a decision is made as to whether a cluster having no dependency either within wafer surface or within die exists) in step S107 as will be detailed later with reference to
If Yes in the step S107, the review apparatus 2 analyzes elements at a representative point and obtains a result of analysis of elements in the defective portion (additional data) in step S108 as will be detailed later with reference to
Following No in the step S107 or following the step S108, the review apparatus 2 ends the process.
Referring now to
In
As shown in
Referring now to
Turning now to
As described above, according to the first process example in the present embodiment, when a defect is found out in the semiconductor device in the course of its manufacturing process, an image of the defective portion and an image of a normal portion corresponding thereto can be obtained as information profitable for post-review. This can improve the possibility that the user can know the cause of the defect.
Next, a second process example will be described by making reference to
Steps S100 to S103, S105, S107 and S108 in
If Yes is determined in the step S103, the SEM type review unit 25 acquires in step S110 an SEM image of a critical dimensioning pattern 51 (see
Subsequently, the optical review unit 24 acquires in step S112 an optical microscopic image of an alignment measurement pattern 53 (see
If Yes is issued in the step S105, the SEM type review unit 25 acquires in step S120 an SEM image of an actual pattern (uncontaminated part of pattern for reference use) nearby the representative defect in cluster and in step S121, captures an SEM image (additional data) of the actual pattern at a site in a given die.
Thereafter, in step S122, the optical review unit 24 captures an optical microscopic image of the actual pattern near the representative defect in cluster and in step S123, captures an optical microscopic image (additional data) of the actual pattern at a site in the given die.
Since, in the steps S120 to S123, the images are captured in the same die, the individual patterns for measurement of, for example, critical dimension are not prepared position by position and the actual pattern is used. In case the objective is a memory product, relatively similar patterns can be found with ease but in other cases, even images of patterns which are not always identical to one another are captured and compared. In that case, it is not necessary to separate images for alignment measurement and thickness measurement and a microscopic image of an actual pattern of each kind is captured.
In the steps S110 and S111, the steps S112 and S113 and the steps S114 and S115, not only images are captured but also measurements may be practiced by having the functions of critical dimensioning measurement, alignment measurement and thickness measurement.
In this manner, according to the second process example in the present embodiment, images for permitting comparison in critical dimension, alignment and thickness between a defective portion and a normal portion corresponding thereto can be acquired and therefore, it is possible to improve the possibility that the user can know the principal cause of a defect or the cause of accelerating degradation of the defect with high speed and high accuracy. By increasing the number of individual images to be captured, a wide range can be watched and the probability of recognizing abnormality (defect) can be raised but the number can be set arbitrarily. In case the number of individual images to be captured is increased, predetermined and specified data which is to be set in the category in the defect information 21 shown in
As set forth so far, with the review apparatus 2 according to the present embodiment, images directly related to the estimation of the cause of a fault can be obtained automatically and hence, the cause of a defect can be searched and clarified more fast and accurately in the present invention than in the case where the person first examines the review result data and again observes the defective portion. Further, in the conventional method in which data of images at the same sites in a defective die and in an adjacent die are acquired, respectively, the difference between the images is relatively small and cannot sometimes become aware of but by capturing images at sites distant from each other (one site being outside of a cluster is the minimum condition. Preferably, the one site is as remote as possible from the cluster) as in the review apparatus 2 of the present embodiment, the difference can be grasped clearly.
In closing giving the description of the present embodiment, it should be understood that the present invention is in no way limited to the embodiment and examples set forth so far. For example, the semiconductor device representing the inspection objective is not limited to the wafer but the present invention may be applied to another inspection objective having a pattern such as a matrix arrangement. Further, the present invention can be modified and altered in specified structure and construction without departing from the gist of the present invention.
It should be further understood by those skilled in the art that although the foregoing description has been made on embodiments of the invention, the invention is not limited thereto and various changes and modifications may be made without departing from the spirit of the invention and the scope of the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
2008-122626 | May 2008 | JP | national |