CN102693924A - Detection system, detection method and detection apparatus - Google Patents

Abstract

According to one embodiment, the invention provides a detection system with a sampling detection portion, a detection interpolation portion, a grade classifying portion and an information manufacture portion. The sampling detection portion obtains the characteristics of elements by sampling detection. The detection interpolation portion calculates the characteristics of undetected elements in an interpolation method. The grade classifying portion generates information related to the element set according to the obtained element characteristic collected by sampling detection and the element characteristic value obtained by the detection interpolation portion. The information manufacture portion generates required information according to the information related to the grades and the element set.

Description

Check system, inspection method and testing fixture

The application speciallys permit out the benefit of priority of hope 2011-64253 number based on the Japan that in advance proposed on March 23rd, 2011, and advocates its interests, and its full content is included in here by reference.

Technical field

Here the example (a plurality of) of explanation all relates to check system, inspection method and testing fixture.

Background technology

Use the unified technology of making a plurality of elements of the same race (for example semiconductor element, magnetic head, semiconductor light-emitting elements etc.) of so-called semiconductor technology (semiconductor fabrication) to be known by us.

In the inspection of using the unified element of making of semiconductor technology, survey sample according to necessity.In such sampling check, perhaps survey sample by a certain number of, perhaps check according to predetermined sampling rule by kind.

But,, therefore only carry out surveying sample the possibility that can not carry out appropriate grade classification according to the sampling rule of being scheduled to by a certain number of sampling check or by kind owing to have deviation in the characteristic value of the element that the unification of use semiconductor technology is made.

Summary of the invention

The problem that the present invention wants to solve is exactly that a kind of check system, inspection method and testing fixture that can carry out appropriate grade classification will be provided.

According to an example, the check system that possesses sampling check portion, checks interpolation portion, grade classification portion and information issuing portion is provided.The characteristic value of the element that obtains through sampling check is collected by sampling check portion.Inspection interpolation portion uses interpolation to ask for the characteristic value that did not carry out the unchecked element of above-mentioned sampling check.The characteristic value of the element that grade classification portion tries to achieve according to the above-mentioned characteristic value of collecting that passes through the element that sampling check obtains with through above-mentioned inspection interpolation portion is made the information relevant with the element group by grade.The information issuing desired information that information issuing portion basis is relevant with the said elements group with above-mentioned grade.

According to said structure, can carry out appropriate grade classification.

Description of drawings

Fig. 1 is the block diagram that is used for illustrating the check system of the 1st example;

Fig. 2 is the sketch map that is used for illustrating the situation of surveying sample by the inspection piece;

Fig. 3 is the schematic graph that is used for illustrating the deviation of the characteristic value on the directions X of A piece of example among Fig. 2;

Fig. 4 is the schematic graph that is used for illustrating the situation of the characteristic value that uses the spline interpolation method to ask for unchecked element;

Fig. 5 is the schematic graph that is used for illustrating the situation of the characteristic value that uses the spline interpolation method to ask for unchecked element;

Fig. 6 is the sketch map that is used for illustrating the situation of the characteristic value that uses the spline interpolation method to ask for unchecked element;

Fig. 7 is the sketch map of the calculating that is used for illustrating probability density p (likelihood);

Fig. 8 for be used for illustrating use prediction distribution (t ' | a) infer the schematic graph of the situation of the probability that becomes desirable grade;

Fig. 9 is the flow chart that is used for illustrating the effect of grade classification portion;

Figure 10 A～10E is the sketch map that is used for illustrating the effect of grade classification portion;

Figure 11 is the flow chart that is used for illustrating the effect of sampling condition enactment portion;

Figure 12 A～12D is the signal process chart that is used for illustrating the effect of sampling condition enactment portion;

Figure 13 is the flow chart that is used for illustrating the effect of checking the piece cutting part;

Figure 14 A～14C is the signal process chart that is used for illustrating the effect of checking the piece cutting part;

Figure 15 is the flow chart that is used for illustrating the inspection method of the 2nd example;

Figure 16 is the block diagram that is used for illustrating the testing fixture of the 3rd example.

Embodiment

More (a plurality of) example is described with reference to the accompanying drawings.In the accompanying drawings, same Reference numeral is represented identical or similar part.

With reference to Fig. 1 the 1st example is described.

Fig. 1 is the block diagram that is used for illustrating the check system of the 1st example.

As shown in Figure 1, be provided with sampling check portion 2, inspection interpolation portion 3, grade classification portion 4, information issuing portion 5, sampling condition enactment portion 6 and inspection piece cutting part 7 in the check system 1.

The characteristic value of the element that obtains through sampling check is collected by sampling check portion 2.

Characteristic value through surveying sample the element that obtains can be collected from sampling check device 100.

Sampling check device 100 according to predetermined sampling rule or after the information that provides from sampling condition enactment portion 6, inspection piece cutting part 7 stated survey sample.

For example, in sampling check, can check by the inspection piece after cutting apart using the unified element divisions of making a plurality of of the same race of semiconductor technology to become predetermined a plurality of inspection piece.

Fig. 2 is the sketch map that is used for illustrating the situation of for example in semiconductor wafer, surveying sample by the inspection piece.

In addition, the inspection piece after the A～H among the figure, J, K represent to cut apart, the point among the figure are represented sampling point (position of taking a sample).And, the both direction that the arrow X among the figure, Y represent mutually orthogonal.

Sampling point both can be used as the point that an element is carried out, and also can be used as the point that the zone that comprises a plurality of elements is carried out.In this case, if as the point that the zone that comprises a plurality of elements is carried out, because therefore a plurality of elements of sampling can improve the inspection precision of sampling check.

As sampling check; Can illustrate the element that for example will be positioned at sampling point downcuts from above-mentioned semiconductor wafer; The perhaps electrical characteristic (for example voltage characteristic or current characteristics etc.) of the element that downcuts of inspection etc. is perhaps checked electrical characteristic etc. in that the element that scales off is assembled into the state in the goods under.

Wherein, for the manufacturing cost or the shortening manufacturing time that reduce element, it is effective that the characteristic check abridged of a part of element is surveyed sample.And, in this sampling check, can carry out characteristic check, the judgement very denying or grade classification by the inspection piece according to predetermined sampling rule.

But, use the characteristic value of the unified element of making of semiconductor technology that deviation is arranged.

Fig. 3 is the characteristic value that is used for illustrating on the directions X of the A piece that Fig. 2 illustrated crosses---the schematic graph of the deviation of output voltage values for example.

Image pattern 3 illustrated are such; Because the characteristic value of element has deviation; If therefore, the danger of the reliability reduction of grade important in the production management is arranged then with the grade that for example determines to check all elements that comprise in the piece uniformly through the shared ratio of each grade of sampling check decision.For example; Result in sampling check is under the many situation of the shared ratio of S level product; If the grade that will check all elements that comprise in the piece, then has the possibility that makes the reliability reduction of the grade of paying because of the mixed proportion of other grades (for example A level or B level etc.) all as the S level.

Therefore, pay the form of grade if adopt the deviation of consideration characteristics value not, then in the process that element is assembled in the goods, have produce in the production management or the managerial problem of rate of finished products maybe.And, because the grade difference of element makes the surcharge of the goods of having assembled element different, thus the handle of element grade is held in the cost management, also be important in acceptance rate or the duration management.

Therefore, ask for the characteristic value (presumed value) of the unchecked element that does not carry out the oversampling inspection in this example.

Inspection interpolation portion 3 use interpolations are asked for the characteristic value of the unchecked element that does not carry out the oversampling inspection.

Wherein, under the situation of using the unified manufacturing of semiconductor technology element, can think that the characteristic value of element changes continuously.

Therefore, can ask for the characteristic value of unchecked element according to the characteristic value of the element that is checked through through sampling check.That is, can enough interpolations obtain the element surveyed sample and the characteristic value of the unchecked element between the element.

When in this case, having the characteristic value of a part of element of trying to achieve to differ widely through sampling check.For example, under the situation owing to generation damages in a part of element such as static, the characteristic value of this element differs widely.Therefore, if ask for not the characteristic value of the element that detects, the possibility that becomes big that deviates from of the characteristic value obtained and actual characteristic value is arranged then according to the characteristic value of such element.

Therefore, can before the characteristic value of asking for unchecked element with interpolation, the information of inappropriate characteristic value be removed.

The judgement that whether is inappropriate characteristic value can for example be carried out with mean value and standard deviation.For example, can be with the characteristic value in the scope that does not have entering " mean value ± n standard deviation " as inappropriate characteristic value.In addition, n can get positive integer, can be according to the stability of semiconductor technology, the appropriate changes such as kind of element.

If ask for the characteristic value of unchecked element after so inappropriate characteristic value being removed with interpolation, the characteristic value that then can suppress to obtain becomes big with actual deviating from of characteristic value.

As interpolation, can use for example spline interpolation method (the 1st interpolation).

Fig. 4 is the schematic graph that is used for illustrating the situation of the characteristic value that uses the spline interpolation method to ask for unchecked element.In addition, Fig. 4 is the situation of the characteristic value of the unchecked element on the directions X of asking in any inspection piece that Fig. 2 illustrated crosses.And, longitudinal axis characterization value, transverse axis representes to check the position of element on the directions X in the piece.X1～X6 is the position that becomes each element of sampling check object.

As Fig. 4 is illustrational, at first use the mean value of characteristic value and the characteristic value of the element become the sampling check object to ask for interpolation function Sj1 (x)～Sj5 (x) by the interval through piecewise polynomial approximation method etc.

Equally, each coordinate of Y direction is asked for the interpolation function on the directions X.For example, each row of Y direction is asked for the interpolation function on the directions X.

Then, ask for interpolation function on the Y direction.

Fig. 5 is used for illustrating the schematic graph of situation of asking for the characteristic value of unchecked element with the spline interpolation method.In addition, Fig. 5 is a situation of asking for the characteristic value of the unchecked element on the Y direction in the inspection piece that Fig. 4 illustrated crosses.And, transverse axis characterization value, the longitudinal axis representes to check the position of element on the interior Y direction of piece.Y1～Y6 is the position that becomes each element of sampling check object.

As Fig. 5 is illustrational, at first use the mean value of characteristic value and the characteristic value of the element become the sampling check object to ask for interpolation function Sj1 (y)～Sj5 (y) by each interval through piecewise polynomial approximation method etc.

Equally, each coordinate of directions X is asked for the interpolation function on the Y direction.For example, each row of directions X are asked for the interpolation function on the Y direction.

Then, ask for the interpolation function of respectively checking in the piece.

In addition, for the element of having obtained the interpolation function on some directions, need not to obtain the interpolation function on another direction.For example, for the element of having obtained the interpolation function on the directions X, need not to obtain the interpolation function on the Y direction.

Through using the interpolation function of trying to achieve like this, can obtain the element surveyed sample and the characteristic value of the unchecked element between the element.And, can carry out paying of grade according to the characteristic value of obtaining.

That is, can infer the characteristic value or the grade of unchecked element.

When asking for the characteristic value of unchecked element, if the situation of using the spline interpolation rule to have the precision of inferring to reduce.

Fig. 6 is used for illustrating the sketch map of situation of asking for the characteristic value of unchecked element with the spline interpolation method.In addition, the element survey sample of " ● " among figure expression, " zero " representes the characteristic value (characteristic value of reality) of unchecked element.And the line that connects " ● " is represented the characteristic value that uses the spline interpolation method to obtain.

If use the spline interpolation method to ask for the characteristic value of unchecked element, then as X13～X16 of Fig. 6 representes, what the characteristic value obtained and actual characteristic value were arranged deviates from change big the time.Therefore, in the 2nd interpolation of following narration, ask for the characteristic value of unchecked element as follows.

In the 2nd interpolation; At first ask for the distribution (below be called " prior distribution ") of the characteristic value of unchecked element, from the prior distribution obtained and from the characteristic value that the probability density p (likelihood) that the characteristic value of the element surveyed sample is obtained obtains unchecked element distribute (below be called " posteriority distribution ") according to past information.Then, distribute according to the posteriority of the characteristic value of the unchecked element of obtaining and derive the information (prediction distribution) relevant with the characteristic value that can obtain unchecked element.

Past information can provide from preserving over the preservation portion 10 of Fig. 1 of information.In addition, relevant information in the past in detail after state.

Here illustrated in greater detail the 2nd interpolation again.

In the 2nd interpolation, the element A that surveyed sample _iCharacteristic value a _i(i=1,2 ..., n) and unchecked elements T ' characteristic value t ' be assumed to mutual independent, defer to same normal distribution (average value mu, variances sigma ²).

The characteristic value of unchecked element can be obtained with following step.

At first, calculate probability density p (likelihood) from the characteristic value of the element of the sampling check that is positioned at unchecked periphery.

Fig. 7 is the sketch map of the calculating that is used for illustrating probability density p (likelihood).

Wherein, the element A that surveyed sample _iCharacteristic value be a _i(i=1,2 ..., probability density n) is because supposition a _iDefer to average value mu, variances sigma ²Normal distribution, therefore can use following formula (1) expression.

$\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(a_i-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\·\·\·\left(1\right)$

So under the situation of the value that the characteristic value of the element that is positioned at unchecked elements T ' periphery is crossed for Fig. 7 illustrated, probability density (likelihood) can use following formula (2) to calculate.

$p=\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="HDA0000141748720000041.png,HDA0000141748720000051.png"\; state="\{\{state\}\}">p</figure-callout>}1\&times;\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="HDA0000141748720000011.png,HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}2\&times;\mathrm{<figure-callout\; id="3"\; label="p"\; filenames="HDA0000141748720000011.png,HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}3\&times;\mathrm{<figure-callout\; id="4"\; label="p"\; filenames="HDA0000141748720000011.png,HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}4\&times;\mathrm{<figure-callout\; id="5"\; label="p"\; filenames="HDA0000141748720000011.png,HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}5\&times;\mathrm{<figure-callout\; id="6"\; label="p"\; filenames="HDA0000141748720000051.png,HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}6\&times;\mathrm{<figure-callout\; id="7"\; label="p"\; filenames="HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}7\&times;\mathrm{<figure-callout\; id="8"\; label="p"\; filenames="HDA0000141748720000101.png"\; state="\{\{state\}\}">p</figure-callout>}8$

$=\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(85-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(90-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(92-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(91-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}$

$\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(86-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(93-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(89-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&times;\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{\frac{-{(87-\mathrm{\&mu;})}^2}{2{\mathrm{\&sigma;}}^2}}\&CenterDot;\&CenterDot;\&CenterDot;\left(2\right)$

And, obtain prior distribution according to past information.

As past information, the information that can obtain when for example carrying out complete inspection in the past, the information that has obtained when having checked the goods of having assembled element in the past etc. are as an example.In this case, the past information under can position identical according to kind, element machine identical, testing fixture the identical situation is obtained prior distribution.

If position identical with unchecked elements T ' kind, element is identical, the machine components identical T of testing fixture _j(j=1,2 ..., characteristic value t n) _jMutually independently, defer to mean value θ, variances sigma ²Normal distribution, then unchecked elements T ' prior distribution w (θ, the σ of characteristic value t ' ²) can use following formula (3) expression.

$w(\mathrm{\&theta;},{\mathrm{\&sigma;}}^2)\&Proportional;{\left({\mathrm{\&sigma;}}^2\right)}^{-(\frac{v_0+1}{2}+1)}{e}^{-\frac{{\mathrm{\&lambda;}}_0+n_0{(\mathrm{\&theta;}-{\mathrm{\&mu;}}_0)}^2}{2{\mathrm{\&sigma;}}^2}}\&CenterDot;\&CenterDot;\&CenterDot;\left(3\right)$

Wherein, μ ₀Be characteristic value t _jMean value, n ₀Be imaginary sampling size, v ₀Be the degree of freedom of contrary card side's distribution (inverse chi-square distribution), λ ₀Scale parameter for the distribution of distribution against card side.μ ₀, v ₀And λ ₀According to elements T _j(j=1,2 ..., characteristic value t m) _j, sample number m and characteristic value t _jMeasuring reliability just determine.For example, when to characteristic value t _jMeasuring reliability when high, μ ₀=m, v ₀=m-1, λ ₀=(m+1) * (characteristic value t _jVariance).

Then, from prior distribution w (θ, σ ²) with probability density p (likelihood) long-pending obtain the characteristic value of unchecked element posteriority distribution w (θ, σ | a) as a=(a ₁, a ₂..., a _n).

In this case, posteriority distribution w (θ, σ | a) can use following formula (4) expression.

$w(\mathrm{\&theta;},\mathrm{\&sigma;}|a)={\left({\mathrm{\&sigma;}}^2\right)}^{-(\frac{{\mathrm{<figure-callout\; id="1"\; label="v"\; filenames="HDA0000141748720000041.png,HDA0000141748720000051.png"\; state="\{\{state\}\}">v</figure-callout>}}_1+1}{2}+1)}{e}^{-\frac{{\mathrm{\&lambda;}}_1+n_1{(\mathrm{\&theta;}-{\mathrm{\&mu;}}_1)}^2}{2{\mathrm{\&sigma;}}^2}}\&CenterDot;\&CenterDot;\&CenterDot;\left(4\right)$

But, v ₁, n ₁, λ ₁, μ ₁With following various the expression.

v ₁＝v ₀+m

n ₁＝n ₀+m

${\mathrm{\&lambda;}}_1={\mathrm{\&lambda;}}_0+[\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{(t_j-\stackrel{\&OverBar;}{t})}^2+\frac{{({\mathrm{\&mu;}}_0-\stackrel{\&OverBar;}{t})}^2}{(1/n_0)+(1/n)}]$

${\mathrm{\&mu;}}_1=\frac{n_0{\mathrm{\&mu;}}_0+\mathrm{nt}}{n_0+n}$

Then, from probability density p (likelihood) and posteriority distribution w (θ, σ | a) obtain unchecked elements T ' characteristic value t ' prediction distribution w (t ' | a).

Prediction distribution w (t ' | a) can use following formula (5) expression.

$w\left(t^{\&prime;}\right|a)=\&Integral;p\left(t^{\&prime;}\right|\mathrm{\&theta;},\mathrm{\&sigma;})w(\mathrm{\&theta;},\mathrm{\&sigma;}|a)\mathrm{d\&theta;d\&sigma;}$ …(5)

$=\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}}{e}^{-\frac{{(t^{\&prime;}-\mathrm{\&theta;})}^2}{2{\mathrm{\&sigma;}}^2}}{\left({\mathrm{\&sigma;}}^2\right)}^{-(\frac{{\mathrm{<figure-callout\; id="1"\; label="v"\; filenames="HDA0000141748720000041.png,HDA0000141748720000051.png"\; state="\{\{state\}\}">v</figure-callout>}}_1+1}{2}+1)}{e}^{-\frac{{\mathrm{\&lambda;}}_1+n_1{(\mathrm{\&theta;}-{\mathrm{\&mu;}}_1)}^2}{2{\mathrm{\&sigma;}}^2}}\mathrm{d\&theta;d\&sigma;}$

Because the prediction distribution w that tries to achieve like this (t ' | a) for relevant obtain unchecked elements T ' the information of characteristic value, therefore infer unchecked elements T ' characteristic value be the probability that becomes desirable grade.

Fig. 8 for be used for illustrating use prediction distribution w (t ' | a) infer the schematic graph of the situation of the probability that becomes desirable grade.

In addition, the characteristic value of transverse axis is divided into the grade (S level, A level, B level, C level) of regulation.

In this case, the prediction distribution w that comprises in can enough desirable rate ranges (t ' | the area of the part that curve surrounded a) account for prediction distribution w (t ' | the ratio of the area of the part that curve surrounded a) is inferred the probability that becomes desirable grade.

Under the situation of example shown in Figure 8, become area that the probability of S level can enough dash areas account for prediction distribution w (t ' | the ratio of the area of the part that curve surrounded a) is inferred.For example, under the situation of example shown in Figure 8, can be estimated as, the probability that becomes the S level is 60%, and the probability that becomes the A level is 30%, and the probability that becomes the C level is 10%.

And, can with the grade that becomes maximum probability be estimated as unchecked elements T ' grade.

So, according to the 2nd interpolation, can infer the characteristic value or the grade of unchecked element.And, can infer the probability that becomes desirable grade.

Therefore, owing to can compile by the unchecked element that the probability that become desirable grade is high, therefore can improve the reliability of the grade of being paid etc.

The characteristic value of the element that the grade classification portion 4 of Fig. 1 obtains according to the characteristic value of collecting that passes through the element that sampling check obtains with inspection interpolation portion 3 is made the information relevant with the element group by grade.

Following as an example, the element that illustrates calculated the probability that becomes desirable grade with the 2nd interpolation carries out grade classification.

Fig. 9 is the flow chart that is used for illustrating the effect of grade classification portion 4.

Figure 10 is the sketch map that is used for illustrating the effect of grade classification portion 4.

As shown in Figure 9, at first, the relevant information (step S1) that becomes the probability of desirable grade is provided from the inspection interpolation portion 3 of Fig. 1.

Then, the isocontour order of stating after the selection drafting (step S2).

For example, can be by hierarchal order---promptly begin to be docile and obedient the contour of stating after the drafting from the high element of quality.

Then, make the contour of the probability that becomes desirable grade, the element additional marking (step S3) nearest to the center in territory, abscission zone.

For example shown in Figure 10 A, make the contour of the probability that becomes desirable grade, the element nearest (the for example element of symbol ★) additional marking to the center in territory, abscission zone.

Then, select to attach the highest element (the 1st element) (step S4) of probability that becomes desirable grade in the tagged element.

For example, be selected to the highest element (element of the symbol ★ shown in Figure 10 B) of probability of desirable grade.

That is, be selected to the highest element of probability of desirable grade according to the characteristic value of the element of collecting that passes through the sampling check acquisition with the characteristic value of checking the element that interpolation portion 3 obtains.

Then, judge that probability that selected element becomes desirable grade is whether more than setting (step S5).

Under the situation of the probability that becomes desirable grade, next grade is carried out above-mentioned steps less than setting.The probability that in all grades, becomes desirable grade all under the situation less than setting, can carry out all product examinations.

Then, select to be positioned near the highest element (the 2nd element) of probability that selected element is, become desirable grade, form a group (step S6).

That is, select to be positioned near the highest element of probability that the element of choosing is, become desirable grade, make information about the element group that comprises initial selected element and next selected element.

At for example such element that is positioned at the right side shown in Figure 10 C is to become under the situation of the highest element of the probability of desirable grade, selects this element.

Then, such making shown in the image pattern 10D about comprising the information of the element group of attaching tagged element and selected element.

Then, select to be arranged in the highest element (the 3rd element) of probability that near the element of said elements group becomes desirable grade, be added to (step S7) in the group.

That is, select to be positioned near the high element of probability that the said elements group is, become desirable grade, make the relevant information that also comprises the element group of this element.

For example shown in Figure 10 E, select to be arranged in the highest element of probability that near the element of said elements group becomes desirable grade, join in the group and go.

Then, in the judgment component group quantity of element whether more than setting (step S8).

Bale packing unit when for example, judging whether to reach outbound.

Whether the mean value of then, judging the probability that becomes desirable grade in the said elements group is less than the value of stipulating (step S9).

The mean value that then, will become the probability of desirable grade is about to element less than the value of regulation as a group (step S10).

For the element in proper range is added in the element group as much as possible, the mean value that will become the probability of desirable grade is about to element less than the value of defined as a group.

Then, judge whether to have carried out above-mentioned steps (step S11) to attaching tagged all elements.

Then, judge whether all grades have been carried out above-mentioned steps (step S12).Then, to information issuing portion 5 grade and relevant group information (step S13) are provided.

In addition, the element that all grades do not belong to is checked individually, paid grade according to check result.

And, can infer that sneaked into should be as the danger of the element of other grades.

Promptly; The characteristic value of collecting from the sampling check portion 2 with Fig. 1 that passes through the element that sampling check obtains and ask for the mean value and the standard deviation of characteristic value with the characteristic value of checking the element that interpolation portion 3 obtains can ask for according to the mean value of characteristic value and standard deviation that sneak into should be as the danger of the element of other grades.

For example, in the scope of " mean value ± standard deviation of characteristic value ", exist under the situation of the cut off value between the grade, can be estimated as sneak into should be as the element of other grades dangerous big.

And, can make with based on the relevant information of bale packing classification about the information of danger.For example, can be according to the information of sneaking into the bale packing that be constructed for as the degree of the danger of the element of other grades classifying.

In this case, can with about the information of danger or with the information issuing portion 5 that also offers Fig. 1 based on the relevant information of bale packing classification about the information of danger.

In addition, state in the use under the situation of the 1st interpolation also and can infer that sneak into should be as the danger of the element of other grades from the probability that becomes desirable grade.

If adopt grade classification portion 4, can be suitably and carry out the grade classification of unchecked element effectively.And, can suppress to have divided the deviation of the characteristic value of the element that comprises in the group of grade.Therefore, can improve the reliability of paying to the grade of the group of having divided grade.And, can reduce in order to pay the quantity of the element that grade need check.

Grade and the information issuing desired information (with the bale packing that be used for outbound relevant information) relevant that the information issuing portion 5 of Fig. 1 provides according to grade classification portion 4 with the element group.

And, provide with based on situation about the relevant information of the bale packing classification of the information of danger under, can defer to this information bale packing is classified.And, can the information about danger be added on the bale packing.

In addition, carrying out all product examinations, paying according to this check result under the situation of grade, can be used for the bale packing of shipment according to these grades.

And; Under bale packing situation of difficult based on the grade of paying; Can define the piece that bale packing is used in advance, decision is included in the grade of the element in the piece that this bale packing uses, and perhaps considers to sneak into the grade of the piece that use as this bale packing of decisions such as danger of the element of other grades.

The grade of the piece of in this case, grades maximum in the grade of paying the element that comprises in the piece of using to bale packing being used as bale packing.And, with the number of degrees value of paying the element that comprises in the piece of using to bale packing, obtain its mean value, the grade that the piece that the grade of the mean value that approaches to try to achieve is most used as bale packing relates to.

So, can be used for the bale packing of outbound according to the grade of decision.

And information issuing portion 5 can write down the ID (Identity Document, documentation of identity) of bale packing unit.And can write down the shipment date etc.

When surveying sample, can consider merely characteristic by a certain number of inspection element.

But, under the situation of the deviation that produces characteristic value because of technological procedure change etc.,, then the condition of poor of inferring that produces characteristic value is arranged in the zone that characteristic value changes greatly if change greatly zone and little zone with identical sampling interval inspection characteristic value.And, change little zone at characteristic value and just become tediously long inspection, in the inspection distribution in man-hour, produce waste.And, part classification is being become under the situation of several grades according to characteristic value, owing to do not know necessity of each characteristic value, as long as know that the attribute of relevant grade is just passable, therefore too much inspection also becomes unnecessary.

Therefore, the sampling condition enactment portion 6 of Fig. 1 be that representative is checked with the part (for example delegation) of inspection subject area, appropriateization of carrying out the sampling point setting according to the variation of the characteristic value of trying to achieve through the inspection of a part of inspection subject area.

Figure 11 is the flow chart that is used for illustrating the effect of sampling condition enactment portion 6.

Figure 12 is the signal process chart that is used for illustrating the effect of sampling condition enactment portion 6.

As example among Fig. 2, sampling check can be undertaken by the inspection piece.In this case, each checks situation about also having in the piece because of the deviation of generation characteristic values such as process variations.And, have the differentiated situation of deviation of characteristic value on directions X in Fig. 2 and the Y direction.

Under these circumstances, on the big direction of the deviation of characteristic value of each inspection piece, can be that representative is checked with a part.

Here as an example, illustrating on the directions X of certain inspection piece with a part is the situation that representative is checked.

At first, shown in figure 11, extract whole elements (step S21) of the lastrow of the inspection piece shown in Figure 12 A out.

That is, on the directions X of the topmost of checking piece, extract whole elements out.

Then, the characteristic value of whole elements that inspection is extracted out is asked for the variation (contour curve) (step S22) of the characteristic value shown in Figure 12 B.

Then, disturb and remove and smoothing, obtain the contour curve (step S23) shown in Figure 12 C.

Remove in interference---when the information that is about to inappropriate characteristic value is removed, can use mean value and standard deviation.For example, can the interior characteristic value of scope that not have to get into " mean value ± n standard deviation " be removed as inappropriate characteristic value.In addition, n can get positive integer, can be according to the stability of semiconductor technology, the appropriate changes such as kind of element.

Smoothing can for example use the method for moving average to carry out.

Then, shown in figure 11, carry out first order differential and handle (step S24).That is, extract the flex point shown in the figure of upside of Figure 12 D out through first order differential.

Then, confirm the zone (high resample area H) (step S25) of value more than setting that the first order differential processing is calculated.

Then, in the resample area H of the high value of the expression of confirming, concentration (number of sampling of the per unit area) N1 of such sampling point with regulation sets sampling point (step S26) shown in the figure of the downside of Figure 12 D.If it is big that first order differential is handled the value of calculating, then the variation of characteristic value is big.Therefore, handle the concentration of the sampling point in the zone of value more than setting of calculating, can the suppression characteristic value infer bad generation through improving first order differential.

Then, such shown in the figure of image pattern 12D downside, in the zone that is not the resample area H of the high value of expression, use the concentration N2 of the sampling point of regulation to set sampling point (step S27).

If it is little that first order differential is handled the value of calculating, then the variation of characteristic value is little.Therefore, handle the value calculated concentration, can suppress tediously long inspection less than the sampling point in the zone of setting through reducing first order differential.

So, the variation of sampling condition enactment portion 6 first order differential treatment characteristic values is handled the value of calculating according to first order differential and is carried out the setting of sampling point.

In addition, before the sampling point in the resample area H that sets the high value of expression, can not the setting of the sampling point in the zone of resample area H of the high value of expression.And, can begin to carry out successively the setting of sampling point from desirable direction.

In this case, the concentration N1 of sampling point, N2 can confirm according to the stability of semiconductor technology, the kind of element etc. in advance.

Set the sampling point (step S28) in this inspection piece like this.

Then, likewise set other and check the sampling point (step S29) in pieces.

The information of the relevant sampling point of setting like this offers the sampling check device 100 of Fig. 1 and surveys sample.

If adopt sampling condition enactment portion 6, can carry out the setting of appropriate sampling point.Therefore, bad generation not only can be suppressed to check, and the appropriate inspection distribution in man-hour can be carried out.

As stated, sampling check can be checked each the inspection piece after cutting apart using the unified element divisions of making a plurality of of the same race of semiconductor technology to become predetermined a plurality of inspection piece.And the quantity that can make the element that inspection comprises in the piece is that equal extent ground will check that subject area cuts apart uniformly.

But, be that equal extent ground will check that subject area cuts apart uniformly if make the quantity of the element that inspection comprises in the piece, then existence can not be sought the possibility of appropriateization of variance of characteristic value.That is, if the quantity of the element that comprises in the piece with inspection is that benchmark is cut apart the inspection subject area, the possibility that then exists the variance of characteristic value in each inspection piece to differ widely.

Therefore, the inspection piece cutting part 7 of Fig. 1 makes the variance of the characteristic value of each piece check appropriateization that subject area is cut apart with diminishing.As the index of the variance of the characteristic value of estimating each piece, for example both can be the maximum of the variance of characteristic value, also can be the difference etc. of the variance of characteristic value.

Figure 13 is the flow chart that is used for illustrating the effect of checking piece cutting part 7.

Figure 14 is the signal process chart that is used for illustrating the effect of checking piece cutting part 7.

At first, shown in figure 13, will check that subject area is divided into m * n (nm) piece (step S31).

In addition, m is the columns on the directions X, and n is the line number on the Y direction.

In this case, cut apart, lift 8 timesharing such as grade (m=4, n=2) that are divided into such shown in Figure 14 A here and describe for example though can wait.

Then, such shown in Figure 14 B, ask for the variance (σ nm, ij [k=1～nm]) (step S32) of characteristic value by each piece.

In addition, characteristic value can adopt through the characteristic value of the element checked and the characteristic value of unchecked element.And, can also adopt the characteristic value when carrying out all product examinations.

Then, ask for the average value mu nm of the variance of characteristic value, ij (step S33).

Then, change and cut apart number, obtain and respectively cut apart several maximum σ of the variance of characteristic value down _MaxNm, ij (step S34).

For example, such maximum σ that obtains the variance of respectively cutting apart the characteristic value under several shown in the image pattern 14C _MaxNm, ij.

In this case, become many, the maximum σ of the variance of characteristic value if cut apart number _MaxNm, ij are decrescence.

Then, obtain the maximum σ of the variance of characteristic value _MaxNm, the reduction rate of ij is less than counting (step S35) the cutting apart of minimum of setting.

For example, obtain the maximum σ of the variance of characteristic value _MaxNm, the reduction rate of ij is less than the number of cutting apart of 0.5% minimum.

In this case, reduction rate can be according to the stability of semiconductor technology, the appropriate changes such as kind of element.

That is, such shown in the image pattern 14C, from cutting apart the maximum σ of the variance under several nm _MaxNm, ij is to the maximum σ of cutting apart the variance under several nm+1 _MaxThe reduction rate that nm+1, ij reduce is less than under 0.5% the situation, and cutting apart several nm is the number of asking for of cutting apart.

Then, cut apart the conduct a survey information (step S36) of piece of numeral system according to what try to achieve.

The information of the inspection piece of making like this offers sampling check device 100 and surveys sample.

Appropriateization of the inspection piece that inspection piece cutting part 7 carries out both can be carried out according to necessity, also can carry out termly.

Below illustrate the inspection method of the 2nd example.

In addition, in following illustrational each step, about the item identical with the effect of above-mentioned check system 1 suitably omitted detailed explanation.

Figure 15 is the flow chart that is used for illustrating the inspection method of the 2nd example.

At first, carry out the sampling check (step S41) of the characteristic value of element according to predetermined sampling rule.

Then, obtain the characteristic value (step S42) that did not carry out the unchecked element of above-mentioned sampling check with interpolation.

In this case, can ask for the characteristic value of unchecked element according to the characteristic value of the element that is checked through through sampling check.That is, can enough interpolations obtain the element surveyed sample and the characteristic value of the unchecked element between the element.For example, can obtain the characteristic value of unchecked element with above-mentioned the 1st interpolation or the 2nd interpolation.

In addition, about the 1st interpolation or the 2nd interpolation because with above-mentioned identical, therefore detailed explanation is omitted.

Then, form element group (step S43) according to the characteristic value of the element of surveying sample and the characteristic value of the unchecked element of trying to achieve by grade.

In this case, the step that can for example defer to example shown in Figure 9 is carried out grade classification.

And situation about illustrating with effect with grade classification portion 4 is the same, can obtain the danger that sneak into as the element of other grades.

And, can make and the relevant information of bale packing classification based on the information of associated danger property.

Then, according to the information issuing desired information (step S44) relevant with the said elements group with grade.

For example, be constructed for being used for the information of the bale packing of shipment.

The inspection method of this example is the same with the effect of the sampling condition enactment portion 6 of above-mentioned Fig. 1, can carry out appropriateization (step S51) that sampling point is set.

That is, can be that representative is checked with the part (for example delegation) of inspection subject area, carry out appropriateization that sampling point is set according to the variation of the characteristic value of obtaining through the part inspection of inspection subject area.

In addition, appropriateization of the sampling point step that for example can defer to example shown in Figure 11 is carried out.

And the inspection method of this example is the same with the effect of the inspection piece cutting part 7 of above-mentioned Fig. 1, can check appropriateization (step S52) that subject area is cut apart.That is, can make the variance of the characteristic value of each inspection piece check appropriateization that subject area is cut apart with diminishing.

In addition, appropriateization cut apart of the inspection subject area step of deferring to example shown in Figure 13 is carried out.

Illustrate the testing fixture of the 3rd example below.

The computer of the above-mentioned inspection method of the testing fixture enough execution of ability of the 3rd example constitutes.

Figure 16 is the block diagram that is used for illustrating as the computer system of the testing fixture of the 3rd example.

Shown in figure 16, be provided with computer 201, input part 202, display part 203 and sampling check preservation as a result portion 204 in the computer system 200.

Be provided with in the computer 201 the calculating part 201a that carries out various information processings, temporarily preserve information the interim preservation 201b of portion such as RAM (RandomAccess Memory, random access memory), control information transmitting-receiving the 201c of input and output portion, preserve the 201d of preservation portion etc. of the audit program of the 3rd example.

In addition, owing to can known technology be used for the each part mentioned above that computer 201 is provided with, so these detailed explanations are omitted.

In order to carry out a series of inspection method, audit program is kept among the 201d of preservation portion that is arranged in the computer 201.Audit program can for example offer computer 201 with the state that is kept in the storage medium, is saved among the 201d of preservation portion that is provided with in the computer 201 through reading.In addition, audit program also can pass through LAN (Local Area Network, local area network (LAN)) etc. and is saved among the 201d of preservation portion that is provided with in the computer 201.

And the audit program that is kept among the 201d of preservation portion is read among the interim preservation 201b of portion, in calculating part 201a, carries out various calculating.At this moment, necessary information is from input part 202 inputs, and result of calculation etc. can be presented in the display part 203 according to necessity.And, be connected with sampling check preservation as a result portion 204 on the input part 202.The result of the sampling check of carrying out according to predetermined sampling rule is kept in the sampling check preservation as a result portion 204.

In this case, the audit program of each step below the execution is kept among the 201d of preservation portion.The computer 201 that preserve this audit program, can carry out these each steps can be regarded the device that possesses respectively with each corresponding unit of step as.

(1) collection is through the step of the characteristic value of the element of sampling check acquisition.

In this case, can collect the result of sampling check through the 201c of input and output portion from sampling check preservation as a result portion 204.

(2) the use interpolation calculates the step of the characteristic value of the unchecked element that does not carry out the oversampling inspection.

The characteristic value of the element that passes through the sampling check acquisition that (3) basis is collected and the characteristic value of the unchecked element that calculates are by the step of the rating calculation information relevant with the element group.

(4) step of the output of the execution information relevant with the element group with grade.

(5) carry out the step of appropriateization that sampling point sets.

(6) check the step of appropriateization that subject area is cut apart.

In addition and since the content of each step can with cross in above-mentioned check system or inspection method illustrated identical, therefore detailed explanation is omitted.

And the testing fixture of the 3rd example both can have been deferred to said sequence time series ground and carry out, and also can carry out and carry out side by side or selectively in time series ground.

And the testing fixture of the 3rd example both can be the device of handling with single calculating part, also can be the device that carries out dispersion treatment with a plurality of calculating parts.

According to each example of above explanation, can realize to carry out check system, inspection method and the testing fixture of suitable grade classification.

Although clear several examples of the present invention, but these examples are the forms that propose as an example, do not limit the intention of scope of invention.These new examples can be implemented with other various forms, in not exceeding the scope of inventing aim, can carry out various omissions, replacement or change.These examples or its distortion not only are included in scope of invention or the aim, are also contained in the invention of putting down in writing in the claim scope and its impartial scope.

Above-mentioned each example can mutual combination be implemented.

And the configuration of each key element that for example above-mentioned check system possessed, quantity etc. are not limited to illustrate, and can suitably change.

Claims (11)

1. a check system is characterized in that,

Possess:

The characteristic value through the element of sampling check acquisition is collected by sampling check portion;

Inspection interpolation portion uses interpolation to ask for the characteristic value that did not carry out the unchecked element of above-mentioned sampling check;

Grade classification portion, the characteristic value of the element of trying to achieve according to the above-mentioned characteristic value of collecting that passes through the element that sampling check obtains with through above-mentioned inspection interpolation portion is made the information relevant with the element group by grade; And

Information issuing portion is according to the information issuing desired information relevant with the said elements group with above-mentioned grade.

2. check system as claimed in claim 1; Also possesses sampling condition enactment portion; This sampling condition enactment portion checks the part of inspection subject area; The variation of the characteristic value of the element of trying to achieve according to the inspection of the part through above-mentioned inspection subject area, appropriateization of carrying out the setting of sampling point.

3. check system as claimed in claim 1,

Above-mentioned sampling check is by checking that the inspection piece after subject area is cut apart carries out;

Also possess the variance that makes the above-mentioned characteristic value in each above-mentioned inspection piece and carry out the inspection piece cutting part of appropriateization of cutting apart of above-mentioned inspection subject area with diminishing.

4. check system as claimed in claim 1, above-mentioned inspection interpolation portion uses the spline interpolation method to ask for the characteristic value of above-mentioned unchecked element.

5. check system as claimed in claim 1,

Above-mentioned inspection interpolation portion asks for the prior distribution of the characteristic value of above-mentioned unchecked element according to past information;

According to above-mentioned prior distribution and the probability density obtained from the characteristic value of the element of above-mentioned sampling check, the posteriority of asking for the characteristic value of above-mentioned unchecked element distributes;

Ask for the probability density of the characteristic value of above-mentioned unchecked element according to above-mentioned posteriority distribution.

6. check system as claimed in claim 1,

The characteristic value of the element that above-mentioned grade classification portion tries to achieve according to the above-mentioned characteristic value of collecting that passes through the element that sampling check obtains with through above-mentioned inspection interpolation portion is selected to the 1st the highest element of probability of desirable grade;

Selection is positioned near the 2nd high element of probability that above-mentioned the 1st element is, become desirable grade, makes and the relevant information of said elements group that comprises above-mentioned the 1st element and above-mentioned the 2nd element;

Selection is positioned near the 3rd high element of probability that the said elements group is, become desirable grade;

Make and the relevant information of said elements group that also comprises above-mentioned the 3rd element.

7. check system as claimed in claim 1,

The characteristic value of the element that above-mentioned grade classification portion tries to achieve according to the above-mentioned characteristic value of collecting that passes through the element that sampling check obtains with through above-mentioned inspection interpolation portion is asked for the mean value and the standard deviation of characteristic value;

Ask for according to the mean value of above-mentioned characteristic value and above-mentioned standard deviation also that sneak into should be as the danger of the element of other grades.

8. check system as claimed in claim 1, sampling condition enactment portion first order differential is handled the variation of above-mentioned characteristic value, carries out the setting of sampling point according to handling the calculated value of trying to achieve through above-mentioned first order differential.

9. inspection method,

Possess:

Carry out the step of sampling check of the characteristic value of element according to sampling rule;

Use interpolation to ask for the step of the characteristic value of the unchecked element that did not carry out above-mentioned sampling check;

Form the step of element group by grade according to the characteristic value of the characteristic value of the element of above-mentioned sampling check and above-mentioned unchecked element of trying to achieve; And

Step according to the information issuing desired information relevant with the said elements group with above-mentioned grade.

10. testing fixture possesses:

Preserve the preservation portion of the characteristic value of the element that obtains through sampling check;

According to the above-mentioned characteristic value that is kept in the above-mentioned preservation portion; Use interpolation to calculate the characteristic value that did not carry out the unchecked element of above-mentioned sampling check; And according to the characteristic value of the above-mentioned element of collecting that passes through the sampling check acquisition and the characteristic value of the unchecked element that aforementioned calculation goes out, by the calculating part of the rating calculation information relevant with the element group;

Execution is from the efferent of the output of the information relevant with the said elements group with above-mentioned grade of aforementioned calculation portion acquisition; And

The display part that shows the information relevant with the said elements group with above-mentioned grade.

11. testing fixture as claimed in claim 10, aforementioned calculation portion and efferent are by computer realization.

Images