CN102693924B - 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 goes out the benefit of priority of No. 2011-64253rd, hope based on the Japanese Patent that on March 23rd, 2011 proposes in advance, and advocates its interests, and its full content is included in here by reference.

Technical field

Here the example (multiple) illustrated all relates to check system, inspection method and testing fixture.

Background technology

Use the technology of so-called semiconductor technology (semiconductor fabrication) the multiple those elements of unified manufacture (such as semiconductor element, magnetic head, semiconductor light-emitting elements etc.) known to us.

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

But, there is deviation owing to using in the characteristic value of the unified element manufactured of semiconductor technology, therefore only carry out by a certain number of sampling check or carry out surveying sample the possibility can not carrying out appropriate grade classification by kind according to predetermined sampling rule.

Summary of the invention

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

According to an example, provide the check system possessing sampling check portion, check interpolation portion, grade classification portion and information preparing department.The characteristic value of the element obtained by sampling check is collected in sampling check portion.Check that interpolation portion uses interpolation to ask for the characteristic value of the unchecked element not carrying out above-mentioned sampling check.Grade classification portion, according to the characteristic value of the above-mentioned element obtained by sampling check collected and the characteristic value of element of being tried to achieve by above-mentioned inspection interpolation portion, makes the information relevant with element group by grade.Information desired by information preparing department makes according to the information relevant with said elements group with above-mentioned grade.

According to said structure, appropriate grade classification can be carried out.

Accompanying drawing explanation

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

Fig. 2 is for illustrating by checking that block carries out the schematic diagram of the situation surveyed sample;

Fig. 3 is the schematic graph of the deviation of characteristic value in the X-direction of A block for illustrating example in Fig. 2;

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

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

Fig. 6 is for illustrating the schematic diagram using spline interpolation method to ask for the situation of the characteristic value of unchecked element;

Fig. 7 is for being used for the schematic diagram of the calculating illustrating probability density p (likelihood);

Fig. 8 for be used for illustrating usage forecastings distribution (t ' | a) presumption becomes the schematic graph of the situation of the probability of desired grade;

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

Figure 10 A ~ 10E is the schematic diagram of the effect for illustrating grade classification portion;

Figure 11 is the flow chart of the effect for illustrating sampling condition configuration part;

Figure 12 A ~ 12D is the signal process chart of the effect for illustrating sampling condition configuration part;

Figure 13 is the flow chart for illustrating the effect checking block comminute portion;

Figure 14 A ~ 14C is the signal process chart for illustrating the effect checking block comminute portion;

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

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

Embodiment

More (multiple) example is described with reference to the accompanying drawings.In the accompanying drawings, same Reference numeral represents same or similar part.

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

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

As shown in Figure 1, be provided with sampling check portion 2, inspection interpolation portion 3, grade classification portion 4, information preparing department 5, sampling condition configuration part 6 in check system 1 and check block comminute portion 7.

The characteristic value of the element obtained by sampling check is collected in sampling check portion 2.

The characteristic value of the element obtained by sampling check can be collected from sampling check device 100.

Sampling check device 100 according to predetermined sampling rule or described later from sampling condition configuration part 6, check that the information that provides of block comminute portion 7 is surveyed sample.

Such as, in sampling check, predetermined multiple inspection block can be divided into by using the unified multiple those elements manufactured of semiconductor technology, check by the inspection block after segmentation.

Fig. 2 is for illustrating such as in the semiconductor wafer by checking that block carries out the schematic diagram of the situation surveyed sample.

In addition, the A ~ H in figure, J, K represent the inspection block after segmentation, and the point in figure represents sampling point (carrying out the position sampled).Further, arrow X, Y in figure represent mutually orthogonal both direction.

Sampling point both can as the point carried out an element, also can as the point carried out the region comprising multiple element.In this case, if as the point carried out the region comprising multiple element, owing to sampling multiple element, therefore, it is possible to improve the inspection precision of sampling check.

As sampling check, can illustrate and such as the element being positioned at sampling point be cut from above-mentioned semiconductor wafer, or check the electrical characteristic (such as voltage characteristic or current characteristics etc.) etc. of element cut, or check electrical characteristic etc. being assembled into by the element scaled off under the state in goods.

Wherein, in order to reduce the manufacturing cost of element or shorten manufacturing time, be effective by the sampling check of the characteristic check abridged of a part of element.Further, in this sampling check, characteristic check can be carried out by inspection block, carry out good no judgement or grade classification according to predetermined sampling rule.

But, use the characteristic value of the unified element manufactured of semiconductor technology to have deviation.

Fig. 3 is the characteristic value in the X-direction for illustrating the A block illustrated in Fig. 2---the schematic graph of the deviation of such as output voltage values.

As illustrating in Fig. 3, because the characteristic value of element has deviation, if therefore determine by the ratio shared by each grade such as determined by sampling check the grade checking all elements comprised in block uniformly, then there is the danger that the reliability of grade important in production management reduces.Such as, when the ratio of result shared by S level product surveyed sample is many, if using the grade of all elements that comprises in this inspection block all as S level, then what have the mixed proportion because of other grades (such as A level or B level etc.) to make the reliability of the grade of paying reduce may.

Therefore, if adopt the deviation not considering characteristic value and the form of paying grade, then element is being assembled in the process in goods the possibility having the problem produced in production management or on yield management.Further, because the grade difference of element makes the surcharge of the goods assembling element different, therefore component level is also important being held in cost management, in acceptance rate or duration management.

Therefore, the characteristic value (presumed value) of the unchecked element not carrying out oversampling inspection is asked in this example.

Check that interpolation portion 3 uses interpolation to ask for not carry out the characteristic value of the unchecked element of oversampling inspection.

Wherein, when using semiconductor technology unified manufacture element, can think that the characteristic value of element changes continuously.

Therefore, it is possible to ask for the characteristic value of unchecked element according to the characteristic value of the element by being checked through.That is, the characteristic value of the unchecked element between element and element surveyed sample can be obtained with interpolation.

In this case, when having the characteristic value by surveying sample a part of element of trying to achieve to differ widely.Such as, when producing damage due to electrostatic etc. in a part of element, the characteristic value of this element differs widely.Therefore, if ask for the characteristic value of the element do not detected according to the characteristic value of such element, then there is the characteristic value obtained to become large possibility with actual deviating from of characteristic value.

Therefore, it is possible to the information of inappropriate characteristic value is removed before the characteristic value asking for unchecked element with interpolation.

Be whether that the judgement of inappropriate characteristic value can such as be carried out with mean value and standard deviation.Such as, can using the characteristic value in the scope not entering " mean value ± n standard deviation " as inappropriate characteristic value.In addition, n can get positive integer, can according to appropriate changes such as the kinds of the stability of semiconductor technology, element.

If so the later interpolation of inappropriate characteristic value removing to be asked for the characteristic value of unchecked element, then the characteristic value obtained can be suppressed to become large with actual deviating from of characteristic value.

As interpolation, such as spline interpolation method (the 1st interpolation) can be used.

Fig. 4 is for illustrating the schematic graph using spline interpolation method to ask for the situation of the characteristic value of unchecked element.In addition, Fig. 4 is the situation of the characteristic value of the unchecked element asked in the X-direction in any inspection block of illustrating in Fig. 2.Further, longitudinal axis characterization value, transverse axis represents the position of element in the X-direction in inspection block.X1 ~ X6 is the position of each element becoming sampling check object.

As Fig. 4 is illustrational, the characteristic value of element becoming and survey sample object and the mean value of characteristic value is first used to ask for interpolation function Sj1 (x) ~ Sj5 (x) by piecewise polynomial approximation method etc. by interval.

Equally, each coordinate of Y-direction is asked for the interpolation function in X-direction.Such as, each row of Y-direction is asked for the interpolation function in X-direction.

Then, the interpolation function in Y-direction is asked for.

Fig. 5 is the schematic graph for illustrating the situation asking for the characteristic value of unchecked element by spline interpolation method.In addition, Fig. 5 is the situation of the characteristic value of the unchecked element asked in the inspection block that illustrated in Fig. 4 in Y-direction.Further, transverse axis characterization value, the longitudinal axis represents the position checking element in Y-direction in block.Y1 ~ Y6 is the position of each element becoming sampling check object.

As Fig. 5 is illustrational, the characteristic value of element becoming and survey sample object and the mean value of characteristic value is first used to ask for interpolation function Sj1 (y) ~ Sj5 (y) by piecewise polynomial approximation method etc. by each interval.

Equally, each coordinate of X-direction is asked for the interpolation function in Y-direction.Such as, each row of X-direction are asked for the interpolation function in Y-direction.

Then, the interpolation function in each inspection block is asked for.

In addition, for the element of the interpolation function obtained on some directions, without the need to obtaining the interpolation function on another direction.Such as, for the element of the interpolation function obtained in X-direction, without the need to obtaining the interpolation function in Y-direction.

By using the interpolation function of trying to achieve like this, the characteristic value of the unchecked element between element and element surveyed sample can be obtained.Further, paying of grade can be carried out according to the characteristic value obtained.

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

When the characteristic value asking for unchecked element, if use spline interpolation rule to have the situation of presumption precision reduction.

Fig. 6 is the schematic diagram for illustrating the situation asking for the characteristic value of unchecked element by spline interpolation method.In addition, the "●" in figure represents the element surveyed sample, and "○" represents the characteristic value (actual characteristic value) of unchecked element.Further, the line connecting "●" represents the characteristic value using spline interpolation method to obtain.

If use spline interpolation method to ask for the characteristic value of unchecked element, then as represented in the X13 ~ X16 of Fig. 6, what have a characteristic value of characteristic value and the reality obtained deviates from when becoming large.Therefore, in following the 2nd interpolation described, the characteristic value of unchecked element is asked for as follows.

In the 2nd interpolation, first ask for the distribution (hereinafter referred to as " prior distribution ") of the characteristic value of unchecked element according to past information, obtain characteristic value distribution (hereinafter referred to as " Posterior distrbutionp ") of unchecked element from the prior distribution obtained and the probability density p (likelihood) that obtains from the characteristic value of the element surveyed sample.Then, the information (prediction distribution) relevant with the characteristic value that can obtain unchecked element is derived according to the Posterior distrbutionp of the characteristic value of the unchecked element obtained.

Past information can provide from preserving over the storage unit 10 of Fig. 1 of information.In addition, being described in detail later about past information.

Here illustrated in greater detail the 2nd interpolation again.

In the 2nd interpolation, the element A 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 by following steps.

First, probability density p (likelihood) is calculated from the characteristic value of element of the sampling check being positioned at unchecked component periphery.

Fig. 7 is for being used for the schematic diagram of the calculating illustrating probability density p (likelihood).

Wherein, the element A surveyed sample _icharacteristic value be a _i(i=1,2 ..., probability density n) is due to supposition a _idefer to average value mu, variances sigma ²normal distribution, therefore, it is possible to represent with following formula (1).

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

So, when the characteristic value of the element being arranged in unchecked elements T ' periphery be Fig. 7 illustrated value, probability density (likelihood) can calculate with following formula (2).

$p=p1\×p2\×p3\×p4\×p5\×p6\×p7\×p8$

$=\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(85-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(90-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(92-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(91-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}$

$\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(86-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(93-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(89-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\×\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{\frac{-{(87-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}}\·\·\·\left(2\right)$

Further, prior distribution is obtained according to past information.

As past information, can by the information obtained when such as all checking in the past, checked the goods assembling element in the past time the information that obtains etc. exemplarily.In this case, can position that is identical according to kind, element machine that is identical, testing fixture identical when past information obtain prior distribution.

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

$w(\mathrm{\θ},{\mathrm{\σ}}^2)\∝{\left({\mathrm{\σ}}^2\right)}^{-(\frac{v_0+1}{2}+1)}{e}^{-\frac{{\mathrm{\λ}}_0+n_0{(\mathrm{\θ}-{\mathrm{\μ}}_0)}^2}{2{\mathrm{\σ}}^2}}\·\·\·\left(3\right)$

Wherein, μ ₀for characteristic value t _jmean value, n ₀for imaginary sampling size, v ₀for the degree of freedom of inverse card side's distribution (inverse chi-square distribution), λ ₀for the scale parameter of the distribution of inverse card side distribution.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 _jmeasurement reliability height determine.Such as, when to characteristic value t _jthe reliability height of measurement time, 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 Posterior distrbutionp w (θ, σ | a) as a=(a ₁, a ₂..., a _n).

In this case, Posterior distrbutionp w (θ, σ | a) can represent with following formula (4).

$w(\mathrm{\θ},\mathrm{\σ}|a)={\left({\mathrm{\σ}}^2\right)}^{-(\frac{v_1+1}{2}+1)}{e}^{-\frac{{\mathrm{\λ}}_1+n_1{(\mathrm{\θ}-{\mathrm{\μ}}_1)}^2}{2{\mathrm{\σ}}^2}}\·\·\·\left(4\right)$

But, v ₁, n ₁, λ ₁, μ ₁variously to represent with following.

v ₁＝v ₀+m

n ₁＝n ₀+m

${\mathrm{\λ}}_1={\mathrm{\λ}}_0+[\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{(t_j-\stackrel{\‾}{t})}^2+\frac{{({\mathrm{\μ}}_0-\stackrel{\‾}{t})}^2}{(1/n_0)+(1/n)}]$

${\mathrm{\μ}}_1=\frac{n_0{\mathrm{\μ}}_0+\mathrm{nt}}{n_0+n}$

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

Prediction distribution w (t ' | a) can represent with following formula (5).

$w\left(t^{\′}\right|a)=\∫p\left(t^{\′}\right|\mathrm{\θ},\mathrm{\σ})w(\mathrm{\θ},\mathrm{\σ}|a)\mathrm{d\θd\σ}$ …(5)

$=\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-\frac{{(t^{\′}-\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2}}{\left({\mathrm{\σ}}^2\right)}^{-(\frac{v_1+1}{2}+1)}{e}^{-\frac{{\mathrm{\λ}}_1+n_1{(\mathrm{\θ}-{\mathrm{\μ}}_1)}^2}{2{\mathrm{\σ}}^2}}\mathrm{d\θd\σ}$

Due to try to achieve like this prediction distribution w (t ' | a) for the unchecked elements T of relevant acquisition ' the information of characteristic value, therefore estimate unchecked elements T ' characteristic value be the probability becoming desired grade.

Fig. 8 for be used for illustrating usage forecastings distribution w (t ' | a) presumption becomes the schematic graph of the situation of the probability of desired 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, can with the prediction distribution w comprised in desired rate range (t ' | the area of the part that Curves a) surrounds account for prediction distribution w (t ' | the ratio presumption of the area of the part that Curves a) surrounds becomes the probability of desired grade.

When exemplified by Fig. 8, become S level probability can with the area of dash area account for prediction distribution w (t ' | the ratio presumption of the area of the part that Curves a) surrounds.Such as, when exemplified by Fig. 8, can be estimated as, the probability becoming S level is 60%, and the probability becoming A level is 30%, and the probability becoming C level is 10%.

Further, the grade becoming maximum probability can be estimated as unchecked elements T ' grade.

So, according to the 2nd interpolation, characteristic value or the grade of unchecked element can be estimated.And, the probability of the grade becoming desired can be estimated.

Therefore, owing to unchecked element high for the probability becoming desired grade can be collected, therefore, it is possible to improve the reliability etc. of the grade of paying.

The grade classification portion 4 of Fig. 1 makes with element group relevant information with the characteristic value of the element obtained with inspection interpolation portion 3 by grade according to the characteristic value of the element by sampling check acquisition collected.

Following as an example, illustrate that the element to having calculated the probability of the grade becoming desired with the 2nd interpolation carries out grade classification.

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

Figure 10 is the schematic diagram of the effect for illustrating grade classification portion 4.

As shown in Figure 9, first, the information (step S1) of the probability about becoming desired grade is provided from the inspection interpolation portion 3 of Fig. 1.

Then, select to draw isocontour order (step S2) described later.

Such as, can by hierarchal order---namely from the element that quality is high according to along drawing contour described later.

Then, the contour of the probability of the grade becoming desired is made, the element additional marking (step S3) nearest to the center in territory, abscission zone.

Such as shown in Figure 10 A, make the contour of the probability of the grade becoming desired, element (such as the element of the symbol ★) additional marking nearest to the center in territory, abscission zone.

Then, select to addition of in the element of mark the highest element (the 1st element) (step S4) of the probability of the grade becoming desired.

Such as, the highest element of the probability of desired grade (element of the symbol ★ shown in Figure 10 B) is selected to.

That is, the highest element of the probability of desired grade is selected to according to the characteristic value of the element obtained by sampling check collected with the characteristic value of element that inspection interpolation portion 3 obtains.

Then, the element selected by judgement becomes the probability of desired grade whether more than setting (step S5).

When the probability becoming desired grade is less than setting, above-mentioned steps is carried out to next grade.The probability becoming desired grade in all grades is all less than setting, all product examinations can be carried out.

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

That is, select to be positioned near the element chosen, become the highest element of the probability of desired grade, make relevant comprise initial selected element with next selected by the information of element group of element.

Be, when becoming the highest element of the probability of desired grade, select this element at the element being such as positioned at right side shown in Figure 10 C like that.

Then, as shown in Figure 10 D, make the information of the element group about comprising element and the selected element that addition of mark.

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

That is, selecting to be positioned near said elements group, become the high element of the probability of desired grade, making the information of the element group about also comprising this element.

Such as shown in Figure 10 E, select the element be arranged near said elements group to become the highest element of the probability of desired grade, join in group and go.

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

Such as, bale packing unit when reaching outbound is judged whether.

Then, whether the mean value judging to become in said elements group the probability of desired grade is less than the value (step S9) of regulation.

Then, the mean value of the probability becoming desired grade is about to be less than the element of the value of regulation as a group (step S10).

In order to be added in element group as much as possible by the element in proper range, the mean value of the probability becoming desired grade is about to be less than the element of the value of defined as a group.

Then, judge whether to have carried out above-mentioned steps (step S11) to all elements that addition of mark.

Then, judge whether to have carried out above-mentioned steps (step S12) to all grades.Then, grade and the information (step S13) about group is provided to information preparing department 5.

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

Further, can estimate that be mixed into should as the danger of the element of other grades.

Namely, from mean value and the standard deviation of being asked for characteristic value by the characteristic value of the element surveying sample the characteristic value of element that obtains and obtain with inspection interpolation portion 3 collected with the sampling check portion 2 of Fig. 1, can ask for according to the mean value of characteristic value and standard deviation that be mixed into should as the danger of the element of other grades.

Such as, when there is the cut off value between grade in the scope of " mean value ± standard deviation of characteristic value ", can be estimated as that be mixed into should be large as the danger of the element of other grades.

Further, relevant information of classifying with the bale packing based on the information about danger can be made.Such as, can according to being mixed into the information that be constructed for classification bale packing as the degree of the danger of the element of other grades.

In this case, by the information about danger or the information preparing department 5 of Fig. 1 can be also supplied to the relevant information of classifying of the bale packing based on the information about danger.

In addition, also can be mixed into from the probability presumption becoming desired grade when using above-mentioned 1st interpolation should as the danger of the element of other grades.

If adopt grade classification portion 4, can suitably and effectively carry out the grade classification of unchecked element.Further, the deviation of the characteristic value having divided the element comprised in the group of grade can be suppressed.Pay to the reliability of the grade of the group dividing grade therefore, it is possible to improve.Further, the quantity needing to carry out the element checked in order to pay grade can be reduced.

Information (information relevant with the bale packing for outbound) desired by the grade that the information preparing department 5 of Fig. 1 provides according to grade classification portion 4 and the information relevant with element group make.

Further, when providing the information of classifying relevant with the bale packing based on the information about danger, this information can be deferred to bale packing is classified.Further, the information about danger can be added on bale packing.

In addition, when carrying out all product examinations, paying grade according to this check result, the bale packing that can carry out for shipment according to these grades.

And, when the bale packing difficulty based on the grade of paying, the block of bale packing can be pre-defined, determine the grade of the element be included in the block of this bale packing, or consider to be mixed into the grade that determine the block of this bale packing as the danger of the element of other grades etc.

In this case, using the grade of grades maximum in the grade of element that comprises in the block paid to bale packing as the block of bale packing.Further, by the number of degrees value of element comprised in the block paid to bale packing, its mean value is obtained, using the grade that the grade closest to the mean value of trying to achieve relates to as the block of bale packing.

In the bale packing that it is possible to carry out for outbound according to the grade determined.

Further, information preparing department 5 can record the ID (Identity Document, documentation of identity) of bale packing unit.And the shipment date etc. can be recorded.

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

But, when producing the deviation of characteristic value because of technological procedure variation etc., if check that characteristic value changes large region and little region with identical sampling interval, then have the situation that the presumption that produces characteristic value is bad in the characteristic value region changed greatly.Further, change little region at characteristic value and just become tediously long inspection, in the distribution checking man-hour, produce waste.Further, when according to characteristic value part classification being become several grade, owing to not knowing necessity of each characteristic value, as long as know that the attribute about grade is just passable, therefore too much inspection also becomes unnecessary.

Therefore, the sampling condition configuration part 6 of Fig. 1 is with the part in check object region (such as a line) for representative checks, appropriateization of sampling point setting is carried out in the change of the characteristic value of trying to achieve according to the inspection of the part by check object region.

Figure 11 is the flow chart of the effect for illustrating sampling condition configuration part 6.

Figure 12 is the signal process chart of the effect for illustrating sampling condition configuration part 6.

As example in Fig. 2, sampling check can be undertaken by inspection block.In this case, the situation of the deviation because of generation characteristic values such as process variations is also had in each inspection block.Further, X-direction in fig. 2 and Y-direction exists the differentiated situation of deviation of characteristic value.

Under these circumstances, the direction that the deviation of the characteristic value of each inspection block is large can be representative checks with a part.

Here as an example, illustrate that be the situation that representative is carried out checking with a part in the X-direction that certain checks block.

First, as shown in figure 11, whole elements (step S21) of the most lastrow of the inspection block shown in Figure 12 A are extracted out.

That is, the X-direction of topmost checking block extracts whole element out.

Then, check the characteristic value of the whole elements extracted out, ask for change (contour curve) (the step S22) of the characteristic value shown in Figure 12 B.

Then, carry out interference removing and smoothing, obtain the contour curve (step S23) shown in Figure 12 C.

In interference removing---when the information removing by inappropriate characteristic value, mean value and standard deviation can be used.Such as, the characteristic value in the scope not entering " mean value ± n standard deviation " can be removed as inappropriate characteristic value.In addition, n can get positive integer, can according to appropriate changes such as the kinds of the stability of semiconductor technology, element.

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

Then, as shown in figure 11, first differential process (step S24) is carried out.That is, extracted out the flex point shown in figure of the upside of Figure 12 D by first differential.

Then, the region of value more than setting (high resample area H) (step S25) that first differential process calculates is determined.

Then, in the resample area H of the high value of the expression determined, as shown in the figure of the downside of Figure 12 D, sampling point (step S26) is set with concentration (number of sampling of the per unit area) N1 of the sampling point of regulation.If the value that first differential process calculates is large, then the change of characteristic value is large.Therefore, by improving the concentration of the sampling point in the region of value more than setting that first differential process calculates, suppression characteristic value bad generation can be estimated.

Then, shown in the figure on the downside of Figure 12 D, not the region of the resample area H representing high level, setting sampling point (step S27) with the concentration N2 of the sampling point of regulation.

If the value that first differential process calculates is little, then the change of characteristic value is little.Therefore, the value calculated by reducing first differential process is less than the concentration of the sampling point in the region of setting, can suppress tediously long inspection.

So, the change of sampling condition configuration part 6 first differential treatment characteristic value, the value calculated according to first differential process carries out the setting of sampling point.

In addition, represent the sampling point in the resample area H of high value in setting before, it can not be the setting of the sampling point in the region of the resample area H representing high value.Further, from desired direction, the setting of sampling point can be carried out successively.

In this case, concentration N1, the N2 of sampling point can pre-determine according to the kind of the stability of semiconductor technology, element etc.

Sampling point (step S28) in this inspection block of such setting.

Then, the sampling point (step S29) in other inspection blocks is similarly set.

The sampling check device 100 that information about the sampling point of setting is like this supplied to Fig. 1 is surveyed sample.

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

As mentioned above, sampling check can be divided into predetermined multiple inspection block by using the unified multiple those elements manufactured of semiconductor technology, checks each inspection block after segmentation.Further, can make the quantity of the element comprised in inspection block be equal extent check object region is split uniformly.

But, if make the quantity of the element comprised in inspection block be equal extent check object region is split uniformly, then there is the possibility appropriately changed of variance can not seeking characteristic value.That is, if to check that the quantity of the element comprised in block is for benchmark is to split check object region, then the possibility that the variance that there is characteristic value in each inspection block differs widely.

Therefore, the inspection block comminute portion 7 of Fig. 1 makes the variance of the characteristic value of each piece carry out appropriateization of check object Region Segmentation with diminishing.As the index of the variance of the characteristic value of each piece of evaluation, both can be such as 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 for illustrating the effect checking block comminute portion 7.

Figure 14 is the signal process chart for illustrating the effect checking block comminute portion 7.

First, as shown in figure 13, check object Region Segmentation is become m × n (nm) block (step S31).

In addition, m is the columns in X-direction, and n is the line number in Y-direction.

In this case, although can segmentation be waited, lift here and be divided into 8 timesharing such as grade (m=4, n=2) to be described for example like that shown in Figure 14 A.

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

In addition, characteristic value can adopt the characteristic value of the characteristic value by surveying sample checked element and unchecked element.Further, characteristic value when carrying out all product examinations can also be adopted.

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

Then, change Segmentation Number, obtain the maximum σ of the variance of characteristic value under each Segmentation Number _maxnm, ij (step S34).

Such as, as shown in Figure 14 C, obtain the maximum σ of the variance of the characteristic value under each Segmentation Number _maxnm, ij.

In this case, if Segmentation Number becomes many, the maximum σ of the variance of characteristic value _maxnm, ij are decrescence.

Then, the maximum σ of the variance of characteristic value is obtained _maxthe reduction rate of nm, ij is less than the minimum Segmentation Number (step S35) of setting.

Such as, the maximum σ of the variance of characteristic value is obtained _maxthe minimum Segmentation Number that the reduction rate of nm, ij is less than 0.5%.

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

That is, shown in Figure 14 C, at the maximum σ from the variance under Segmentation Number nm _maxnm, ij are to the maximum σ of the variance under Segmentation Number nm+1 _maxwhen the reduction rate that nm+1, ij reduce is less than 0.5%, Segmentation Number nm is the Segmentation Number asked for.

Then, the information (step S36) checking block is made according to the Segmentation Number of trying to achieve.

The information of the inspection block of such making is supplied to sampling check device 100 and surveys sample.

Check that appropriateization of the inspection block that block comminute portion 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 omits detailed description.

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

First, the sampling check (step S41) of the characteristic value of element is carried out according to predetermined sampling rule.

Then, the characteristic value (step S42) of the unchecked element not carrying out above-mentioned sampling check is obtained with interpolation.

In this case, the characteristic value of unchecked element can be asked for according to the characteristic value of the element by being checked through.That is, the characteristic value of the unchecked element between element and element surveyed sample can be obtained with interpolation.Such as, the characteristic value of unchecked element can be obtained with above-mentioned 1st interpolation or the 2nd interpolation.

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

Then, according to sampling check to the characteristic value of element and the characteristic value of unchecked element of trying to achieve by grade forming element group (step S43).

In this case, the step can such as deferred to exemplified by Fig. 9 carries out grade classification.

Further, the same with situation about illustrating with the effect in grade classification portion 4, the danger that be mixed into as the element of other grades can be obtained.

Further, relevant information of classifying with the bale packing of the information based on associated danger can be made.

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

Such as, the information of the bale packing carried out for shipment is constructed for.

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

That is, can with the part in check object region (such as a line) for representative check, appropriateization of sampling point setting is carried out in the change of the characteristic value obtained according to the part inspection by check object region.

In addition, the step that appropriateization of sampling point such as can be deferred to exemplified by Figure 11 is carried out.

Further, the inspection method of this example is the same with the effect in the inspection block comminute portion 7 of above-mentioned Fig. 1, can carry out appropriateization (step S52) of check object Region Segmentation.That is, the variance of the characteristic value of each inspection block can be made with diminishing to carry out appropriateization of check object Region Segmentation.

In addition, the step that appropriateization of check object Region Segmentation defers to Figure 13 example shown is carried out.

Illustrate the testing fixture of the 3rd example below.

The testing fixture of the 3rd example can be formed with the computer performing above-mentioned inspection method.

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

As shown in figure 16, computer 201, input part 202, display part 203 and sampling check result storage unit 204 is provided with in computer system 200.

Be provided with the input and output portion 201c of the transmitting-receivings of storage unit 201b, control information temporarily such as the calculating part 201a performing various information processing, the interim stored RAM of guarantor (RandomAccess Memory, random access memory) in computer 201, preserve the storage unit 201d etc. of the audit program of the 3rd example.

In addition, owing to known technology can be used for each part mentioned above arranged in computer 201, therefore these detailed description are omitted.

In order to perform a series of inspection method, audit program is kept in the storage unit 201d be arranged in computer 201.Audit program such as can be supplied to computer 201 with the state be kept in storage medium, is saved in the storage unit 201d arranged in computer 201 by reading.In addition, audit program also can pass through LAN (Local Area Network, local area network (LAN)) etc. be saved in computer 201 arrange storage unit 201d in.

Further, the audit program be kept in storage unit 201d is read out in interim storage unit 201b, in calculating part 201a, carry out various calculating.Now, necessary information inputs from input part 202, and result of calculation etc. can be presented in display part 203 according to necessity.Further, input part 202 is connected with sampling check result storage unit 204.Be kept in sampling check result storage unit 204 according to the result of the regular sampling check carried out of predetermined sampling.

In this case, the audit program performing following steps is kept in storage unit 201d.Preserve this audit program, the computer 201 of these each steps can be performed can regard the device possessing unit corresponding with each step respectively as.

(1) step of the characteristic value of the element obtained by sampling check is collected.

In this case, the result of sampling check can be collected from sampling check result storage unit 204 by input and output portion 201c.

(2) interpolation is used to calculate the step of not carrying out the characteristic value of the unchecked element of oversampling inspection.

(3) according to the characteristic value of element and the characteristic value of the unchecked element calculated that are obtained by sampling check the collected step by the rating calculation information relevant with element group.

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

(5) step of appropriateization of sampling point setting is carried out.

(6) step of appropriateization of check object Region Segmentation is carried out.

In addition, due to each step content can with illustrated in above-mentioned check system or inspection method identical, therefore detailed description omit.

Further, the testing fixture of the 3rd example both can have been deferred to said sequence and time-sequentially perform, and also can time-sequentially perform and perform side by side or selectively.

Further, the testing fixture of the 3rd example both can be carry out with single calculating part the device that processes, also can be the device carrying out dispersion treatment with multiple calculating part.

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

Although the description of several example of the present invention, but these examples are the forms proposed as an example, do not limit the intention of scope of invention.These new examples can be implemented with other various forms, can carry out various omission, replacement or change in the scope not exceeding invention aim.These examples or its distortion are not only included in scope of invention or aim, are also contained in the scope of invention and its equalization recorded in right.

Above-mentioned each example can combine enforcement mutually.

And the configuration, quantity etc. of each key element that such as above-mentioned check system possesses are not limited to and illustrated, can suitably change.

Claims (10)

1. a check system, is characterized in that,

Possess:

Sampling check portion, collects the characteristic value of the element obtained by sampling check;

Check interpolation portion, use interpolation to ask for the characteristic value of the unchecked element not carrying out above-mentioned sampling check;

Grade classification portion, according to the characteristic value of the above-mentioned element obtained by sampling check collected and the characteristic value of element of being tried to achieve by above-mentioned inspection interpolation portion, makes the information relevant with element group by grade;

Information preparing department, the information desired by making according to the information relevant with said elements group with above-mentioned grade; And

Sampling condition configuration part, checks the part in check object region, according to the inspection of the part by above-mentioned check object region and the change of the characteristic value of the element of trying to achieve, carries out appropriateization of the setting of sampling point.

2. check system as claimed in claim 1,

Above-mentioned sampling check is undertaken by by the inspection block after check object Region Segmentation;

Also possesses the inspection block comminute portion making the variance of the above-mentioned characteristic value in each above-mentioned inspection block carry out appropriateization of the segmentation in above-mentioned check object region with diminishing.

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

4. 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, ask for the Posterior distrbutionp of the characteristic value of above-mentioned unchecked element;

The probability density of the characteristic value of above-mentioned unchecked element is asked for according to above-mentioned Posterior distrbutionp.

5. check system as claimed in claim 1,

Above-mentioned grade classification portion, according to the characteristic value of the above-mentioned element obtained by sampling check collected and the characteristic value of element of being tried to achieve by above-mentioned inspection interpolation portion, is selected to the 1st element that the probability of desired grade is the highest;

Select to be positioned near above-mentioned 1st element, become the 2nd high element of the probability of desired grade, make the information relevant with the said elements group comprising above-mentioned 1st element and above-mentioned 2nd element;

Select to be positioned near said elements group, become the 3rd high element of the probability of desired grade;

Make the information relevant with the said elements group also comprising above-mentioned 3rd element.

6. check system as claimed in claim 1,

Above-mentioned grade classification portion, according to the characteristic value of the above-mentioned element obtained by sampling check collected and the characteristic value of element of being tried to achieve by above-mentioned inspection interpolation portion, asks for mean value and the standard deviation of characteristic value;

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

7. check system as claimed in claim 1, the change of the above-mentioned characteristic value of sampling condition configuration part first differential process, carries out the setting of sampling point according to the calculated value of being tried to achieve by above-mentioned first differential process.

8. an inspection method,

Possess:

The step of the sampling check of the characteristic value of element is carried out according to sampling rule;

Interpolation is used to ask for the step of the characteristic value of the unchecked element not carrying out above-mentioned sampling check;

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

The step of the information desired by making according to the information relevant with said elements group with above-mentioned grade; And

The part in check object region is checked, according to the inspection of the part by above-mentioned check object region and the change of the characteristic value of the element of trying to achieve, carries out the step of appropriateization of the setting of sampling point.

9. a testing fixture, possesses:

Storage unit, preserves the characteristic value of the element obtained by sampling check;

Calculating part, according to the above-mentioned characteristic value be kept in above-mentioned storage unit, interpolation is used to calculate the characteristic value not carrying out the unchecked element of above-mentioned sampling check, and according to above-mentioned the preserved characteristic value of element by sampling check acquisition and the characteristic value of the above-mentioned unchecked element calculated, by the information that rating calculation is relevant with element group;

Efferent, performs the output of the information relevant with said elements group with above-mentioned grade obtained from above-mentioned calculating part;

Display part, shows the information relevant with said elements group with above-mentioned grade; And

Sampling condition configuration part, checks the part in check object region, according to the inspection of the part by above-mentioned check object region and the change of the characteristic value of the element of trying to achieve, carries out appropriateization of the setting of sampling point.

10. testing fixture as claimed in claim 9, above-mentioned calculating part and efferent are realized by computer.