DE10036118A1 - Influence calculation of failure or breakdown signatures with regional yield data e.g. for manufacture of semiconductor devices, involves preparing yield data table containing several regional yield rates for processed wafers - Google Patents

Abstract

An influence calculation method for failure/breakdown signatures through multiple regression with regional yield data, requires initially breaking down the total arrays of IC-devices which are present on a processed wafer by establishing several adjacent zones and then preparing a yield data table in which several yield rates of the process data are contained. Several columns are present which indicated whether relevant wafers are influenced by a given signature, or not i.e. a column for each signature. The zones of each wafer are identified as influenced by the signatures and a mean yield influence of the signature per influenced wafer is calculated, and a total yield influence of the signatures per influence yield area are obtained, in which initially the weighting of the wafer is calculated and influenced by the signature in one yield area by dividing the total number by wafers, which are influenced by the signatures.

Description

The present invention relates generally to one Influence calculation method for failure signatures at regional yield data used for the production of Semiconductor devices and integrated circuits is suitable; in particular, the present invention relates to a Influence of failure signatures by multiple Regression in regional yield data including the Type of influence on yield, by calculating the influence on individual areas, and by calculating the influence on selected subjects.

Complex industrial processes, for example those that are used for Manufacture of semiconductor integrated circuit devices used typically require a few tens to a few hundred tightly controlled individual steps and parameters up for completion. In particular, the manufacturing process includes for semiconductor ICs the modification of the physical Composition and geometry of semiconductor material in complex patterns and in extremely small dimensions in order to the IC to achieve the desired electrical functions. Disruptions and fluctuations in the manufacturing environment  to deformations of the IC structure, which in turn lead to that the chip cannot perform its function. About that In addition, it is usually not possible to test a Semiconductor wafers are not performed until with the wafer go through a large number of critical process steps was immediately to determine at what point in the Process caused a defect in a wafer, or what caused the defective wafers, unless a defect mechanism during the manufacturing process is predominantly dominant. Because a dominant defect mechanism statistically a small sample, it can happen that decisions based on such a small sample are hit, don't address the most essential problems, or that the cause of the defect is not necessarily related to a problem with a certain part of the Manufacturing facility stands. The problem could be Be the result of an operating error. Because the number at Variables that can cause a chip to be a "Failure" is referred to, so large is the attempt extremely troublesome, processed wafers through their respective Machining flow paths using manual identification and track analysis of each failed wafer. Regardless of the cause, a failed device represents is a reduction in process yield and it is the goal a yield improvement team to determine which ones Defects the most important in limiting the yield and then take appropriate measures to ensure that To reduce the occurrence of these defects.

The defect mechanisms that occur in the IC devices formed on a wafer can be roughly divided into two categories: local and total. Overall defect mechanisms affect all of the chips in an area of the wafer, or the entire wafer, in a similar manner, whereas the local defect mechanisms affect small areas of isolated chips independently of other chips on the wafer. Examples of local defects include pinholes, pipes, and photolithography point defects. Manufacturing defects such as mask misalignment, change in process temperature and heating time, and uneven impurity concentration across the wafer are some examples of overall deformation mechanisms. In addition, only those wafers whose defective IC devices form visually identifiable patterns or "signatures" will generally be found at Yield modeling and taken into account in yield measurement methods. In Fig. 1, a wafer 10 is shown, on which several arrays (fields) of IC devices or chips are produced with the aid of a photolithography device, with an identifiable first signature 13 , a second signature 15 and a third signature 17 , each of which dark markings 12 represent a failed chip and each of the empty markings 14 represents a good chip. In addition, each visually or non-visually identifiable signature is given a name for reference values, so that the first signature 13 is a signature of the "VF edge" type, which indicates inadequate polishing of the wafer during a CMP process (chemical mechanical polishing ), the second signature 15 can be referred to as the "Sweden" type signature, and the third signature 17 can be referred to, for example, as the "AA-DT misalignment" type signature.

The influence calculation of hard failure signatures is currently running only in a single exploitation compartment. The conventional  The process is ideally suited to individual signatures sort out when on different wafers occur, and on the condition that the Logging procedure of the signatures is correct.

However, experience shows that one hundred percent Logging failure signatures is a task which is very difficult to accomplish and also too expensive is when you take into account the increased accuracy that you get expected due to this measure. Processed wafers that can not be marked at random, from the whole group the wafer should be excluded before calculating is started because they have a low yield (e.g. below 88%). Another situation occurs when a low yield wafer with only a single signature is marked, although it is more than can have a signature, the entire Loss of yield of this wafer incorrectly the one signature is attributed. In addition, there is a statistical Uncertainty comes into play when sometimes the influence of two or several different signatures that they cannot be separated. This is especially the case with overlapping areas of influence the wafer. It is therefore extremely important to determine the exact cause each signature and the type of influence in relation to each of the signatures exactly.

Therefore, a conventional influence calculation method for failure signatures for regional yield data, which is relevant for the production of semiconductor IC devices, typically comprises the following steps:

• 1. Provision of all relevant yield data in Contemplated wafer in a single yield bin,  for example as the corresponding yield rate, and Verification whether the wafer in question is affected by a certain, identifiable failure signature influenced becomes. For example, there may be five wafers be designated W1 through W5, each have a measured yield rate, and at which the Possibility of using one, both, or neither of two failure signatures S1 and S2 be influenced, the capital letter V the Verification of a particular failure signature indicates as indicated in Table 1 below.

TABLE 1

• 1. Obtaining an influence on yield with respect to each of the influencing signatures using a method of multiple regression analysis, wherein:
  Y _avg - I _S1 .S ₁ - I _S2 .S ₂ = Y _actual (1)
  Furthermore, Y _avg , I _S1 and I _{S2 are} defined as the unknowns, and the values of (S ₁ , S ₂ , Y _actual ) = (0, 0, 95%) and (S ₁ , S ₂ , Y _actual ) = (1, 0, 92%) etc. defined as the requirements. Therefore, five equations are provided, each containing the same group of failure signatures S ₁ , S ₂ as the independent variables, and a current yield rate Y _actual , which is the dependent variable, providing the basis for calculating the values of the three unknowns To form Y _avg , IS ₁ and IS ₂ , with which a fit can be carried out on a three-dimensional graph whose axes S ₁ , S ₂ and Y _{actual run} in the direction X, Y and Z, respectively, in a conventional three-dimensional coordinate system. As shown in FIG. 2, each of the intersection points D ₀ , D ₁ , D ₂ and D _{12 represents} the average yield rate of a wafer, corresponding to the combined action of the failure signatures (S ₁ , S ₂ ) at values such as (0, 0 ), (1, 0), (0, 1) and (1, 1), where "1" denotes the verification of the existence of a failure signature, and "0" denotes the opposite. In addition, a plane P is determined by the best fit of at least three points in the three-dimensional coordinate system, which are determined by the mean yield rates D ₀ , D ₁ , D ₂ and D ₁₂ . Direct algebraic replacement therefore gives (Y _avg , I _SI , I _S2 ) = (96%, 4%, 7%), where I _S1 and I _S2 each represent the mean yield influences of the failure signature S ₁ and S ₂ , as this is shown in Figure 2;
• 2. Receiving a total loss of exploitation, which can be attributed to the corresponding failure signature. Referring to Table 1, therefore, the total yield loss due to the failure signature S ₁ is equal to I _S1 .2 / 5 = 1.6%, because the fail signature S _1, two influenced by the five wafers. On the same basis, the total loss of yield due to the failure signature S ₂ is I _S2 .2 / 5 = 2.8%, since the failure signature S ₁ also affects two of the five wafers.

However, the above is conventional Procedure uncertain when two or more failure signatures fail affect most of the measured wafers, for example the Wafer W4 as shown in Table 1. As a result of statistical uncertainty is the calculated mean Influence of yield and also the total loss of yield unsure which of a particular failure signature is attributable.

If it appears that the measured wafer is only influenced by a failure signature, but with a significantly lower current yield rate Y _actual than for the other wafers that are influenced by the same failure signature, this would mean that other factors, for example, an operator error that affects loss of yield on the particular wafer. The current yield rate Y _actual , which is attributed to the failure signature according to the conventional method, therefore has a systematic error (bias), and the total loss of yield due to this failure signature cannot be calculated exactly.

In view of the above disadvantages an advantage of providing the present invention a more precise influence calculation of failure signatures based on regional yield data Silicon wafer through which the relationship between each  Failure signature and therefore the average yield influences respective areas of the processed wafers clearly identified and can be correlated, resulting in a more accurate Total yield influence calculation of the corresponding Failure signature leads.

Therefore, a first influence calculation method for Provided, in which the entire arrays of IC devices that are formed on a processed wafer were, first in several neighboring areas of influence be divided, for example in the form of concentric Rings, and all processed, collected wafers the same imaginary boundary lines are divided to to define the same corresponding areas of influence. About that In addition, a so-called profile is created for each newly recognized Signature defined, each corresponding area of influence of a wafer that is affected by the signature could be a symbol for tracking all relevant Yield data is assigned.

Then the rate of exploitation of each sphere of influence (sometimes referred to as the regional rate of exploitation) of a processed one Wafers measured, and tabulated on a data sheet, which the table contains an average yield rate of each of the To get areas of influence. In addition, the Table of all regional yield rates of the processed Wafers in all respective yield compartments (Loss of yield compared to the previous subject), instead of just one yield number for each wafer in one single swap pocket, as well as additional columns that indicate whether the wafer has a specific signature is affected or not (one column for each signature).  

Thereupon a method of multiple regression analysis used to address each of the direct regional influences Calculate and maintain signature for each affected wafer. Regional influences caused by the same signature in the same area are then averaged, to get a medium regional influence. To one total regional influence of the corresponding signature in Maintaining the wafer process becomes the middle regional one Yield with the total count of the affected wafers multiplied, and then by the total count of the processed wafer divided in an yield compartment. Different this expresses the mean regional yield weighted that the proportion of affected wafers in the Yield compartment is calculated. Therefore, a Overall yield influence of a failure signature over all Areas of influence are summed up to the loss of yield (or the overall yield impact) due to the signature calculate, and then sort and display on Diagrams and tables for further investigation.

Another advantage of the present invention is the provision of a more precise influence calculation of Failure signatures based on exploitation data processed wafer, creating the relationship between each Failure signature and therefore the effects of yield on the wafers respective exploitation subjects clearly identified and can be correlated, resulting in a more accurate Total yield influence calculation of the respective Failure signature leads.

Therefore, a second influence calculation method is used Provided, in which the influence of one or several failure signatures with the yield rate at least  an exploitation subject is correlated (sometimes as the Designated exploitation rate), with several of the Specialist exploitation rates that occur during a series of wafer processes to be measured, tabulated to have an investigation a method of multiple regression analysis perform. Contain this tabulated trade exploitation data the exploitation rates of all respective exploitation subjects, all the more so the yield loss of the processed wafers compared to the previous exploitation subjects instead of just one Yield number for each wafer in only one Yield compartment. In the tabulated technical exploitation data are additional columns are also provided that indicate whether a particular exploitation compartment is influenced by a signature will or not (one column for each signature) and it will a so-called profile for each newly recognized signature defined, and for the exploitation subjects, which by the Signature could be affected.

Thereafter, the exploitation rate of each of the exploitation subjects measured, which is influenced by at least one signature and the exploitation subjects that are identified be influenced by at least one signature. Then it will be a method of multiple regression analysis used, each of the technical effects of a signature to be calculated and maintained for each affected output compartment. To an average influence of the signature per affected wafer in all yield compartments calculate, all influences on the technical yield are added, which concern the same signature. Then one Overall impact of the signature per affected Output compartment (or per affected item) calculated, where first the weighting of the wafers is calculated by the Signature in all exploitation subjects are influenced,  Division of the total number of wafers by the signature be influenced with the total number of IC devices in all exploitation subjects, and subsequent multiplication of the above-mentioned mean influence on yield the calculated weight. Ideally, the Overall yield influence caused by the second Influence calculation method is calculated with the Total yield influence that by the first Influence calculation method was calculated if the same relevant signature is considered. However, the second influence calculation to be used Overview of the deterioration rates and the reasons for to get a certain signature if more than one Output compartment by the same signature throughout Wafer processes is affected.

There is also another advantage of the present Invention in providing a more precise Influence calculation of failure signatures on the basis both the exploitation data and the regional yield data processed wafer, creating the relationship between each Failure signature and therefore the average yield influences Wafers of respective yield compartments and on respective areas the wafer can be clearly identified and correlated to what for a more precise overall yield influence calculation of the failure signature in question.

Therefore, a third influence calculation method is used Provided, in which the entire arrays of IC devices manufactured on a processed wafer were, first in several neighboring areas of influence be divided, for example in the form of concentric Rings, and all processed wafers that pass through  different exploitation subjects are collected through the same imaginary boundary lines are divided to to define the same corresponding areas of influence. The Yield influence of one or more failure signatures with the exploitation rate of at least one exploitation compartment correlated (sometimes referred to as the rate of exploitation), and with the respective regional exploitation rates of processed wafers, with several exploitation rates and also several regional yield rates for each processed Wafers measured by a group of wafer processes and be tabulated to be examined. The tabulated Technical exploitation data and regional yield data include the Specialist exploitation rates of all associated exploitation subjects or the regional yield rates of each processed wafer, and so on the yield loss of the processed wafers compared to the previous exploitation subjects instead of just one Yield number for each wafer in only one Yield compartment. There are also additional ones in the table Columns are provided that indicate whether a particular Output compartment or the respective areas of influence of a processed wafers can be influenced by a signature or not (one column for each signature). Furthermore becomes a so-called profile for each newly recognized signature defined, each yield compartment and each sphere of influence, that could be affected by the signature, a name is assigned for easy tracking of all relevant yield data.

After that, a method of multiple regression analysis used to direct each of the regional and Specialist effects of a signature per affected wafer or to be calculated and influenced per affected yield compartment receive. Regional influences affected by the same  Signature in the same area will then be affected averaged to obtain a medium regional influence. To have an overall regional and technical impact same signature across all to get different exploitation subjects, the mean regional yield with the total number of affected wafers over the total number of processed Wafer weighted during each of the yield influences in Reference to the same signature is weighted in that with the total number of yield compartments in the wafer processes is divided. Therefore, an overall influence of the yield Signature over all areas of influence or over all Specialist influences are summed up to the loss of yield (or the overall yield impact) due to the specific Failure signature to calculate, then using these sizes a sorting and display in diagrams and tables for further investigations are carried out.

The present invention is therefore advantageous in comparison to the conventional influence calculation method for Failure signatures because they are different types of Influence of yield on respective regional yield data of processed wafer, and the influence of yield a signature on individual areas and also on individual ones selected subjects calculated.

The invention is illustrated below with reference to drawings illustrated embodiments explained in more detail what other goals, characteristics and advantages of the present Invention emerge. It shows:

Figure 1 is a plan view of a processed wafer on which arrays of IC devices are provided with distinct, non-overlapping signatures, which are shown as clusters of failed IC devices in dark color.

FIG. 2 shows a perspective view and a side view of a plane P, which is defined by points D ₀ , D ₁ , D ₂ and D ₁₂ , applied with a best-fit method;

Fig. 3 array of IC devices, which are provided on a machined wafer, divided into a plurality of zones of influence in the form of concentric rings in accordance with the present invention;

Fig. 4 is a table of specialized yield data and regional yield data for each processed wafer, correlated with one or more failure signatures according to the present invention; and

Fig. 5 is a schematic representation of the overall yield impact of different failure signatures.

The present invention provides a more precise one Influence calculation of failure signatures available based on both the exploitation data and the regional Yield data of processed wafers, creating the relationship between each failure signature and therefore the middle one Yield effects on wafers of respective yield compartments and clearly identified on respective areas of the wafers and can be correlated, resulting in a more accurate  Total yield influence calculation of the concerned Failure signature leads.

First embodiment

In Fig. 3, it is shown that arrays of IC devices formed on a processed wafer 20 are divided into areas of influence A, B, C, D and E in the form of concentric rings according to the present invention, each rectangle on represents IC device present on the wafer.

The wafer is then tested and measured to obtain an absolute yield of the IC devices on the wafer. Furthermore, a yield data table as shown in Fig. 4 is provided to log the regional yield data of each wafer affected by one or more signatures and the tray yield data of each tray by the same signatures throughout the wafer processes being affected. As shown in FIG. 4, there is shown an yield data table according to the present invention, which contains regional yield data 30 from a plurality of processed wafers (wafers 1 to 21 ) and also a plurality of correlated signature data 32 . A test and measurement is then carried out with the respective regional yield data 30 of each processed wafer 20 in accordance with the yield rate of a particular area of influence of a processed wafer in a given yield compartment, which tests can include a current / voltage test with direct current in the circuit, a voltage / Current test with alternating current in the circuit, a data write test, etc. In particular, the yield data table shown in FIG. 4 contains yield compartments T1, T2 and T3, so that column YT1_A denotes the measured yield rates of area A of wafers 1 to 21 in compartment T1. Furthermore, column YT2_A denotes the measured yield rates of area A of wafers 1 to 21 in compartment T2, and column YT3_A denotes the measured yield rates of area A of wafers 1 to 21 in compartment T3. As is apparent from FIG. 4, the yield data table further contains signature data 32 , which comprise three columns with different signatures S1, S2 and S3, each of which represents an identifiable signature (for example any of the signatures shown in FIG. 1). The capital letter V, which is present in one of the columns S1, S2 and S3, verifies the presence of a certain failure signature, which could negatively influence the yield rate of the wafer in question. For example, the regional exploitation data 30 in the yield data table shown in FIG. 4 indicate that no identifiable signature can be found on the wafer 1 , whereas the wafer 2 turns out to be influenced by the signatures S2 and S3.

In addition, at least one area and at least one Output compartment, which by the signature in question be influenced, identified because only a small one Share of the total areas and the exploitation subjects by the affected signature is affected. Table 2 shows one Example of the result of the exploitation subjects and respective ones Areas of a processed wafer, which by signatures S1, S2 and S3 can be influenced.  

TABLE 2

As can be seen from Table 2, the signature only appears in the Areas D and E of the wafer on only in the yield bin T1 is identified; a corresponding interpretation applies for the other names in Table 2.

A method of multiple regression analysis is then used to calculate and maintain each of the direct regional and specialist yield influences of a signature per affected wafer or per affected yield compartment. For example, the influence of yield I _{T1_D_S1} on signature S1, which is identified in area D of a processed wafer in yield _pocket T1, can be calculated using the following multiple regression equation (2):

YT1_D _avg - I _{T1_D_S1} .S ₁ = YT1_D _actual (2)

However, as shown in Fig. 4, the regional trade yield data 30 indicated in the yield data table indicate that twenty-one groups of the values (S ₁ , YT1_D _actual ) can be obtained directly from the table, so that the values for YT1_D _avg and I _{T1_D_S1} can be obtained by direct algebraic replacements in equation (2). The calculated value I _{T1_D_S1} represents the influence of the yield of the signature S1, which is identified in the region D of a processed wafer in the yield compartment T1. Correspondingly, the influence of the yield I _{T1_E_S1} can be calculated using the same method. As can be seen from Table 2, the signature S1 only influences the yield rates of T1_D and T1_E, which represent the areas D and E of a processed wafer that is identified in the yield compartment T1; therefore other yield influences which relate to the same signature S1, for example I _{T1_A_S1} , I _{T1_B_S1} , I _{T1_E_S1,.} , , etc. with the exception of I _{T1_D_S1} and I _{T1_E_S1} , all set to zero. Therefore, any regional _{impact on bin yields} I _{bin_region_sig} can be calculated based on a specific signature using the same concept.

The mean yield influence I _{0_S1 of} the signature S1, which influences the yield rate of an individual processed wafer, can then be calculated using the following equation (3):

I _{0_S1} = (C _A I _{T1_A_S1} + C _B I _{T1_B_S1} + C _C T _{T1_C_S1} + C _D I _{T1_D_S1} + C _E I _{T1_E_S1} + C _A I _{T2_A_S1} + C _B T _{T2_B_S1} + _... ) / C _total (3)

Here, C _{X represents} the number of IC areas in the area X of a processed wafer, which are influenced by the signature S1, and C _total denotes the total number of IC devices on the wafer. Each of the regional trade impact on signature S1 is weighted by dividing it by the total number C _total of IC devices, and then summed up to obtain a regional trade impact of signature S1 on a single wafer, I _{0_S1} ,

Thereupon, the entire regional influence on the yield of the signature S1 on wafers during the entire wafer processes is calculated using the following equation (4):

I _{impact_S1} = I _{0_S1} .W _S1 / W _total

Here, W _S1 denotes the total number of wafers that are influenced by the signature S1 during the entire wafer processes, and W _total denotes the total number of wafers that are processed during the entire wafer processes.

This leads to the fact that the total regional effects on the yield of the signatures S2 and S3 can be calculated by using the equations described above, these values then being sorted and displayed on diagrams and tables for further investigations. FIG. 5 is a schematic illustration showing the total yield influences of the failure signatures S1, S2 and S3. Therefore, the signature S2 has the greatest value of the influence of the regional technical yield, so that this must be taken into account immediately in order to correct the loss of yield as a result of this signature, so as to improve the overall yield of the entire wafer processes.

Second embodiment

In contrast to the above Influence calculation method for one or more signatures based on both regional yield data and the Technical data is according to the second embodiment of the present invention an influence calculation method for one or more signatures based only on the regional yield data provided.  

In general, the influence calculation method according to the second embodiment comprises the following steps:

• 1. Subdivide the entire arrays of IC devices that are present on a processed wafer into several adjacent areas of influence, for example in the form of concentric rings, as shown in FIG. 3;
• 2. Provision of a yield data table in which the measured yield rate of each area of influence processed wafer is entered, and additional Columns that indicate whether the wafer is through a particular Signature is affected or not (a column for any signature);
• 3. Identify each of the areas of the wafer covered by the signature can be influenced;
• 4. Use of a multiple method Regression analysis to each of the regional influences calculate a signature per affected wafer and to obtain;
• 5. Calculation of the influence of the signature's yield a single wafer, first weighting each Area in the entire wafer is calculated by dividing the total number of IC devices in each Range by the total number of IC devices on the entire wafer; and then each of the above regional influences mentioned with the calculated Weighting is multiplied; and summation  all weighted regional influences of the relevant signature;
• 6. Obtain an overall regional impact on yield the signature on the wafer over the entire wafer process, by calculating the weighting of the wafers by the Signature in an exploitation compartment are influenced, namely by dividing the total number of wafers that be influenced by the signature by which Total number of IC devices in the yield compartment; and subsequent multiplication of the above Yield influence with the calculated weighting, um to maintain the total regional impact on yield.

Third embodiment

In contrast to the above Influence calculation method for one or more signatures based on both regional yield data and the Technical yield data is according to the third embodiment of the present invention an influence calculation method for one or more signatures based only on the Specialist data provided.

Overall, the influence calculation method according to the third embodiment comprises the following steps:

• 1. measurement of the exploitation rates of the respective exploitation subjects, and identifying the exploitation subjects that are covered by at least one signature can be influenced;
• 2. Provision of a yield data table in which several measured specialist exploitation rates are entered,  which is the average yield rate of the wafers in correspond to the subject of exploitation, and at which additional columns are provided which indicate whether the wafer has a specific signature is influenced or not (one column for each Signature);
• 3. Use of a multiple method Regression analysis to determine each of the yield effects one signature per affected output compartment calculate and maintain;
• 4. Calculation of an average influence of the yield Signature per affected wafer in all Yield fans by adding all Fold yield influences; and
• 5. Obtain an overall impact on the subject Signature per affected output compartment (or per influenced items), the weighting of the Wafer is calculated by the signature in all fields of exploitation are influenced, namely by dividing the total number of wafers by the Signature are influenced by the total number of IC devices in all exploitation compartments; and subsequent multiplication of the above outlined mean yield influence with the calculated weighting.

Therefore, the present invention is compared to conventional influence calculation methods for Failure signatures advantageous in that they different types of yield influence on each  regional yield data of the processed wafers taken into account, and the influence of a signature on the yield individual areas and also on individual, selected subjects calculated.

While the present invention has been accomplished on the basis of Embodiments explained in the above illustrated drawings are shown, however Those skilled in the art will appreciate that the invention is not limited to the embodiments since various changes or modifications in this Let respects do without the essence of the invention departing. The nature and scope of the invention are illustrated by the All of the present application documents are determined and are intended to be encompassed by the appended claims.

Claims (13)

1. Influence calculation method for failure signatures by multiple regression with regional yield data, with the following steps:
Subdivision of the entire arrays of IC devices that are present on a processed wafer by defining several adjacent areas;
Providing a yield data table in which:
several regional yield rates of the processed wafers are contained in all respective yield subjects (loss of yield compared to the previous subject); and
there are multiple columns indicating whether or not wafers in question are affected by a particular signature (one column for each signature);
Identifying the areas of each wafer affected by the signatures;
Using a method of multiple regression analysis to calculate and obtain each of the regional yield influences of a signature per affected wafer;
Calculating an average signature impact per affected wafer, first computing the weight of each area in the entire wafer by dividing the total number of IC devices in each area by the total number of IC devices on the entire wafer; and then multiplying and adding each of the above regional yield influences by the calculated weights to obtain the average yield influence per wafer affected; and
Obtain a total yield impact of the signature per affected yield bin (or per affected item), first computing the weight of the wafers affected by the signature in an yield bin by dividing the total number of wafers affected by the signature the total number of IC devices in the yield compartment; and then multiplying the above-mentioned mean yield influence by the calculated weighting. 2. The method according to claim 1, characterized in that the mean yield influence of the signature per affected wafer is calculated by the following equation:
I _{0_Sig} = (C _Region I _{Region_Sig} ) / C _total ;
in which:
I _{0_Sig denotes} the mean _{effects on} yield of the signature Sig per affected wafer;
C _region denotes the total count of IC devices in each area;
I _{region_Sig denotes} the regional yield influences of the signature Sig per affected IC device; and
C _total denotes the total count of IC devices in a given wafer. 3. The method according to claim 2, characterized in that the total yield influence of a signature per affected yield subjects (or per affected item) is calculated by the following equation:
I _{impact_Sig} = I _{0_Sig} .W _sig / W _total ;
in which
I _{impact_Sig denotes} the total yield impact of a signature per affected exploitation subject (or per affected item);
W _sig denotes the total number of wafers in the yield compartments which are influenced by the signature; and
W _total denotes the total number of wafer counts in the yield compartments. 4. The method according to claim 1, characterized in that the adjacent areas the shape of concentric rings exhibit.   5. Influence calculation method for failure signatures by multiple regression for technical exploitation data, with the following steps:
Measuring the exploitation rates of respective exploitation subjects and identifying the exploitation subjects that are influenced by at least one signature;
Providing a yield data table in which
a plurality of measured specialist yield rates are contained, which correspond to the average yield rate of the wafers in the relevant yield subjects; and
multiple columns are included that indicate whether or not exploitation subjects are affected by a particular signature (one column for each signature);
Using a method of multiple regression analysis to calculate and obtain each of the yield effects of a signature per affected yield subject;
Calculation of an average influence on the yield of the signature per affected wafer in all yield compartments, by adding all the effects on the yield; and
Obtain a total shelf impact of the signature per affected bin (or per affected lot), first computing the weight of the wafers affected by the signature in all yield bins by dividing the total number of wafers affected by the signature. by the total number of IC devices in all yield compartments; and then multiplying the above-mentioned mean yield influence by the calculated weighting. 6. The method according to claim 5, characterized in that the mean influence on the yield of the signature per affected wafer is calculated by the following equation:
I _{0_Sig} = I _{Bin_Sig} ;
in which
I _{0_Sig denotes} the mean _{effects on} yield of the signature Sig per affected wafer; and
I _{Bin_Sig describes} the _{impact of} a signature on the yield of each yield. 7. The method according to claim 6, characterized in that the total yield influence of a signature per affected yield subjects (or per affected item) is calculated by the following equation:
I _{impact_Sig} = I _{0_Sig} .W _sig / W _total ;
in which
I _{impact_Sig denotes} the total yield impact of a signature per affected exploitation subject (or per affected item);
W _Sig denotes the total number of wafers in the yield compartments which are influenced by the signature; and
W _total denotes the total number of wafer counts in the yield compartments. 8. The method according to claim 5, characterized in that the Specialist exploitation rate of each of the exploitation subjects by one DC voltage / DC test in the circuit is tested and measured by a AC / AC test in the circuit, or a data writing test. 9. Influence calculation method for failure signatures through multiple regression for regional technical data, with the following steps:
Subdivision of the entire arrays of IC devices that are present on a processed wafer to define a number of adjacent areas;
Providing a yield data table in which:
several regional exploitation rates of the processed wafers are contained in all respective yield subjects (loss of yield compared to the previous subject); and
multiple columns that indicate whether or not wafers in question are affected by a particular signature (one column for each signature);
Measuring the exploitation rates of respective yield subjects and regional yield rates of respective wafers, and identifying the yield subjects and the wafers, which are influenced by at least one signature;
Using a multiple regression analysis method to calculate and obtain each of the regional yield influences of a signature per affected wafer in a respective yield compartment;
Calculating an average yield impact of the signature per affected wafer, first calculating the weight of each area in the entire wafer by dividing the total number of IC devices in each area by the total number of IC devices on the entire wafer; and then multiplying and adding each of the above-mentioned regional yield effects by the calculated weights to obtain the average yield per wafer affected; and
Obtain a total yield impact of the signature per affected yield bin (or per affected item), first computing the weight of the wafers affected by the signature in an yield bin by dividing the total number of wafers affected by the signature the total number of IC devices in the yield compartment; and then multiplying the above-mentioned mean yield influence by the calculated weighting. 10. The method according to claim 9, characterized in that the mean yield influence of the signature per affected wafer is calculated by the following equation:
I _{0_Sig} = (C _Region I _{Bin_Region_Sig} ) / C _total ;
in which
I _{0_Sig denotes} the mean _{effects on} yield of the signature Sig per affected wafer;
C _region denotes the total count of the IC devices in each area;
I _{Bin_Region_Sig denotes} the regional yield influences of the signature Sig per IC device affected in a respective yield _compartment ; and
C _total denotes the total count of IC devices in a relevant wafer. 11. The method according to claim 10, characterized in that the total yield influence of a signature per affected yield subjects (or per influenced item) is calculated by the following equation:
I _{impact_Sig} = I _{0_Sig} .W _sig / W _total ;
in which
I _{impact_Sig denotes} the total yield impact of a signature per affected exploitation subject (or per affected item);
W _sig denotes the total number of wafers in the yield compartments which are influenced by the signature; and
W _total denotes the total number of wafer counts in the yield compartments. 12. The method according to claim 9, characterized in that the adjacent areas the shape of concentric rings exhibit. 13. The method according to claim 9, characterized in that the Specialist exploitation rate of each of the exploitation subjects by one DC voltage / DC test in the circuit is tested and measured by a AC / AC test in the circuit, or a data writing test.