CN101630534B - Method and device for testing reliability of nonvolatile memories - Google Patents

Method and device for testing reliability of nonvolatile memories Download PDF

Info

Publication number
CN101630534B
CN101630534B CN 200810040767 CN200810040767A CN101630534B CN 101630534 B CN101630534 B CN 101630534B CN 200810040767 CN200810040767 CN 200810040767 CN 200810040767 A CN200810040767 A CN 200810040767A CN 101630534 B CN101630534 B CN 101630534B
Authority
CN
China
Prior art keywords
test
inefficacy
block
blocks
criterion
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN 200810040767
Other languages
Chinese (zh)
Other versions
CN101630534A (en
Inventor
简维廷
郭强
龚斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Semiconductor Manufacturing International Shanghai Corp
Semiconductor Manufacturing International Beijing Corp
Original Assignee
Semiconductor Manufacturing International Shanghai Corp
Semiconductor Manufacturing International Beijing Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Semiconductor Manufacturing International Shanghai Corp, Semiconductor Manufacturing International Beijing Corp filed Critical Semiconductor Manufacturing International Shanghai Corp
Priority to CN 200810040767 priority Critical patent/CN101630534B/en
Publication of CN101630534A publication Critical patent/CN101630534A/en
Application granted granted Critical
Publication of CN101630534B publication Critical patent/CN101630534B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Techniques For Improving Reliability Of Storages (AREA)
  • For Increasing The Reliability Of Semiconductor Memories (AREA)

Abstract

The invention discloses a method for testing reliability of nonvolatile memories, which comprises the following steps: respectively selecting N blocks from each of K nonvolatile memories respectively provided with Nm blocks; respectively executing a test program for T times on the selected K*N blocks to acquire test data; according to the test data, counting a data relation between the frequency for executing the test program and the number of blocks accumulatively failing in tests in the frequency, and drawing a test curve; determining an equivalent failed block number judging standard; reading a failure number corresponding to an executing frequency value in the test curve; and comparing the obtained failure number with the equivalent failed block number judging standard, judging whether the nonvolatile memories are qualified. The invention also provides a device for testing the reliability of the nonvolatile memories. The test method and the test device have the advantages of reducing test time and improving test efficiency under the condition of not losing statistical significance.

Description

The method of testing of reliability of nonvolatile memories and device
[technical field]
The present invention relates to integrated circuit manufacturing and field tests, relate in particular to the method and apparatus of reliability of nonvolatile memories test.
[background technology]
Make the field at integrated circuit, along with the characteristic dimension of integrated circuit constantly reduces, chip integration improves constantly, and traditional production model, process materials, device model and method of testing etc. all face lot of challenges.
Nonvolatile memory is a kind of common SIC (semiconductor integrated circuit), and each nonvolatile memory comprises that several are used to store the block of data.Unusual one of the key link in the manufacture process for the reliability testing of nonvolatile memory.The purpose of reliability of nonvolatile memories test is to investigate the reliability of several nonvolatile memories of a certain batch, by calculating the ratio of the underproof storer of failure testing in this batch nonvolatile memory, judge the reliability of this batch nonvolatile memory.In the prior art, it is Standard test programme that comprises several operations such as reading, wipe, write of establishment that certain nonvolatile memory is carried out the basic skills that failure testing adopts, make tens thousand of times of each block even hundreds thousand of ground in this nonvolatile memory repeatedly carry out this program, investigate ratio value of being above standard whether that the block of inefficacy occurs this nonvolatile memory after carrying out this program of specific times, if the value of not being above standard, it is qualified then this nonvolatile memory to be judged to be, otherwise is defective.
In the prior art, in the test to one batch reliability of nonvolatile memories, need in the product of this batch, choose several nonvolatile memories, to the certain number of times Standard test programme of each onblock executing in each nonvolatile memory of choosing.For example has N to one mThe nonvolatile memory of individual block is at N mIndividual block is carried out respectively T mBehind the inferior test procedure, what blocks check has lost efficacy, if the number of blocks that occurs to lose efficacy is less than inefficacy number of blocks criterion n mThen be qualified, otherwise be defective.
Usually require storer to have at least above storage life tens thousand of times to the application of nonvolatile memory at present, so in reliability testing, the number of times T of each onblock executing test procedure mUsually more than tens thousand of time, and usually have thousands of blocks in each nonvolatile memory, i.e. N mUsually more than 1000, therefore above-mentioned test will expend a large amount of test durations.For example for the flash memory of the 2Gbit capacity of standard, each flash memory has 2000 blocks, to each onblock executing 10 5Inferior test procedure in the case, takes 1 year consuming time nearly altogether to the test of a flash memory.So the long test duration is fully unacceptable for integrated circuit manufacturing enterprise.And simple passing through reduces the number of the nonvolatile memory that is used for test or the number of times of reduction execution test procedure shortens the test duration, will certainly reduce the statistical significance of test result, reduces the cogency of test result.
[summary of the invention]
Technical matters to be solved by this invention is, a kind of method and apparatus for the reliability of nonvolatile memories test is provided, do not lose in the statistical significance situation guaranteeing that test result is compared with the method for prior art, reduce the time of test, improve testing efficiency.
In order to address the above problem, the invention provides a kind of method for the reliability of nonvolatile memories test, described method of testing comprises the steps: to have respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block; K * N the block that is selected carried out respectively test procedure T time, described T * N＜T m* N m, described T mFor standard is carried out number of times; According to test data, the data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy is afterwards also drawn test curve; According to the parameter that adopts in the test, determine to adopt the equivalence inefficacy number of blocks criterion n ' of the method; In test curve, read with standard and carry out time numerical value T mCorresponding inefficacy number n; The inefficacy number n that obtains is compared with equivalent inefficacy number of blocks criterion n ', if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
Described standard is carried out number of times T mMeaning be, at all K * N mIndividual block is carried out respectively T mBehind the inferior test procedure, if the number of blocks that occurs to lose efficacy is less than inefficacy number of blocks criterion n mThen be qualified, otherwise be defective.
As optional technical scheme, described T＞T m
As optional technical scheme, described N is not more than N mOne of percentage, T is not less than T m5 times.
As optional technical scheme, according to all K * N mIn the individual block n is arranged mThe probability of individual block inefficacy should with the K * N that an is selected sample in have the probability of the individual block inefficacy of n ' to equate principle, determine equivalent inefficacy number of blocks criterion n '.Think that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mBe identical, the method for determining equivalent inefficacy number criterion n ' is specially to be found the solution following equation:
Σ i = 0 [ n m / K ] binomial ( i ; N m , P m ) = Σ i = 0 [ n ′ / K ] binomial ( i ; N , P m )
As optional technical scheme, satisfy Weibull distribution relation or lognormal distribution relation drafting test curve according to carrying out between the number of blocks that had lost efficacy under test procedure number of times and the corresponding number of times.
As optional technical scheme, described nonvolatile memory is flash memory.
As optional technical scheme, the scope of the value of described K is 50～100, preferred 77.
The present invention also provides a kind of device for the reliability of nonvolatile memories test, has respectively N to K mThe nonvolatile memory of individual block is tested, and testing standard is at all K * N mIndividual block is carried out respectively T mBehind the inferior test procedure, if the number of blocks that occurs to lose efficacy is less than inefficacy number of blocks criterion n mThen be qualified, otherwise be defective.
Described device comprises: block is chosen the unit, is used for having respectively N at K mChoose N block in each of the nonvolatile memory of individual block; The loop test unit is used for that block is chosen K * N the block that the unit is selected and carries out respectively test procedure T time, described T * N＜T m* N m, described T mFor standard is carried out number of times; The Drawing of Curve unit is used for according to test data, and the data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy is afterwards also drawn test curve; The criterion computing unit is used for installing the parameter that adopts in test according to this, determines equivalent inefficacy number of blocks criterion n '; The unit is got in the failure number eye reading, and the test curve that is used for producing from the Drawing of Curve unit is read with standard and carried out time numerical value T mCorresponding inefficacy number n; Comparison judgment unit, being used for that the inefficacy number n that the unit obtains is got in this failure number eye reading compares with the determined equivalent inefficacy number of blocks criterion n ' of criterion computing unit, if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
As optional technical scheme, described T＞T m
As optional technical scheme, described N is not more than N mOne of percentage, T is not less than T m5 times.
As optional technical scheme, in the described criterion computing unit, according to all K * N mIn the individual block n is arranged mThe probability of individual block inefficacy should with the K * N that an is selected sample in have the probability of the individual block inefficacy of n ' to equate principle, determine equivalent inefficacy number of blocks criterion n '.Calculate equivalent inefficacy number criterion n ' time, think that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mIdentical.The method of determining equivalent inefficacy number criterion n ' is specially to be found the solution following equation:
Σ i = 0 [ n m / K ] binomial ( i ; N m , P m ) = Σ i = 0 [ n ′ / K ] binomial ( i ; N , P m )
As optional technical scheme, in the described Drawing of Curve unit, satisfy the Weibull distribution relation between the number of blocks that had lost efficacy under test procedure number of times and the corresponding number of times or lognormal distribution concerns curve plotting according to carrying out.
As optional technical scheme, described nonvolatile memory is flash memory.
As optional technical scheme, the scope of the value of described K is 50～100, preferred 77.
The invention has the advantages that, selected cycle index and the product of number satisfy T * N＜T m* N m, namely reduced total testing time that need to carry out, thereby reached the purpose that reduces total testing time; According to the parameter that adopts in the test standard of test is revised again, with revised standard test result is judged, can guarantee the reliability of conclusion.
[description of drawings]
Accompanying drawing 1 is depicted as the implementing procedure figure for reliability of nonvolatile memories method of testing embodiment provided by the invention;
Accompanying drawing 2 is depicted as that invention provides is used for reliability of nonvolatile memories method of testing embodiment, the test curve shape schematic diagram that is depicted as according to test result;
Accompanying drawing 3 is depicted as the structural representation for reliability of nonvolatile memories proving installation embodiment provided by the invention.
[embodiment]
Elaborate below in conjunction with the embodiment of accompanying drawing to the method and apparatus for reliability of nonvolatile memories test provided by the invention.
Provide at first by reference to the accompanying drawings the embodiment for the reliability of nonvolatile memories method of testing provided by the invention.
Accompanying drawing 1 is depicted as the implementing procedure figure for reliability of nonvolatile memories method of testing embodiment provided by the invention.
This embodiment comprises the steps: step S10, has respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block; Step S11 carries out respectively test procedure T time to K * N the block that is selected, and obtains test data, and described test data comprises the number of inefficacy block and the number of times of each inefficacy block actual execution test procedure when losing efficacy, described T * N＜T m* N m, described T mFor standard is carried out number of times; Step S12, according to test data, the data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy is afterwards also drawn test curve; Step S13 according to the parameter that adopts in the test, determines to adopt the equivalence inefficacy number of blocks criterion n ' of the method; Step S14 reads in test curve with standard and carries out time numerical value T mCorresponding inefficacy number n; Step S15 compares the inefficacy number n that obtains with equivalent inefficacy number of blocks criterion n ', if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
Wherein the execution sequence of step S13 and step S14 can exchange.
Described nonvolatile memory is flash memory, also can be other storeies with Long-term Preservation data characteristics.
For the flash memory of 2G, the technical standard that often adopts in the industry at present be to select 77 nonvolatile memories to carry out above-mentioned test, and for other nonvolatile memories, the number of selection is not quite similar.In order to guarantee to test representative and the saving test duration, the scope of described K value is 50～100 usually.
Step S10 has respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block.This step is tested for choose a part of block in nonvolatile memory, for example for the nonvolatile memory (N with 2000 blocks m=2000), 100 of can only choose wherein of the method test (N=100).
Step S11 carries out respectively test procedure T time to K * N the block that is selected, and obtains test data, and described test data comprises the number of inefficacy block and the number of times of each inefficacy block actual execution test procedure when losing efficacy, described T * N＜T m* N m, described T mFor standard is carried out number of times.
Described standard is carried out number of times T mMeaning be, at all K * N mIndividual block is carried out respectively T mBehind the inferior test procedure, if the number of blocks that occurs to lose efficacy is less than inefficacy number of blocks criterion n mThen be qualified, otherwise be defective.
Comprise the nonvolatile memory of flash memory since have can the Long-term Preservation data characteristics, when therefore it being tested, various functions such as will reading in, read, wipe it is comprehensively tested, described test procedure comprises reads " 0 ", read " 1 ", wipe, write, the operation such as counter-rotating.
For example, when testing standard be to investigate each block to carry out respectively 1 * 10 5Inferior test procedure (T m=1 * 10 5) the rear situation that inefficacy occurs, then in this embodiment, can carry out 5 * 10 5Inferior (T=1 * 10 5).
The technical conceive of this method is to carry out more frequently test operation in less block, to infer according to the curve that obtains whether flash memory is qualified, therefore in step S11, to reduce total testing time in order reaching, thereby reach time-saving purpose, need to satisfy T * N＜T m* N mThis condition.Above-mentioned for example in, for a nonvolatile memory, 100 blocks (N=100) of choosing are carried out respectively 5 * 10 5Inferior test procedure, i.e. T=5 * 10 5, need altogether to carry out 5 * 10 7Inferior, and use conventional methods, need to be to 2000 block (N m=2000) carry out 1 * 10 5Inferior test (T m=1 * 10 5), need altogether to carry out 2 * 10 8Inferior.Both relatively adopt the described method of this embodiment, and the testing time of required execution only be previous 1/4th, and the corresponding spent test duration also is previous 1/4th, so can save the test duration.
The number of times of performed test procedure is greater than standard testing number of times, i.e. T＞T when testing m, can further improve the order of accuarcy of the method.Carry out 5 * 10 in the example example described above 5Inferior test procedure is greater than standard testing number of times 1 * 10 5Inferior.The advantage of this technical scheme is being, can guarantee standard testing number of times T mBe in the interval that determined by actual testing time T (0, T) within, carry out time numerical value T at follow-up drafting test curve and the standard of reading mAmong the step of corresponding inefficacy number n, because curve is by interval (0, T) testing time in and corresponding accumulative total inefficacy number of blocks obtain by match, therefore this curve is interval (0, T) one section relative other part in more can reflect testing time and the truth of number of blocks Relations Among that adds up to lose efficacy exactly, T mBe in the interval (0, can guarantee reading and T within T) mDuring corresponding inefficacy number n, utilization be that curve is in the interval (0, the comparatively accurately part within T) therefore can further improve the levels of precision of the method.
In order significantly to save the test duration, preferred testing scheme is that described N is not more than N mOne of percentage, T is not less than T m5 times.Such as the above-mentioned example, has the nonvolatile memory (N of 2000 blocks m=2000), test (N=100) for 100 that only choose wherein, the testing time of each block is by 1 * 10 5Inferior test (T m=1 * 10 5) bring up to 5 * 10 5Inferior (T=1 * 10 5).
Step S12, according to the test data that step S11 obtains, the data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy is afterwards also drawn test curve.This step statistics be fail message in all the K * N that is selected blocks.Described " number of blocks of inefficacy occurs accumulative total in the test of this number of times " refers in the situation that all number of blocks sums that inefficacy occur and losing efficacy less than this testing time under this testing time.
For example, the information of test data record is: in selecteed all blocks in this test, when test procedure is carried out to 1 * 10 5In the situation of inferior test, finding has 10 blocks to lose efficacy; When test procedure continues to carry out to 2 * 10 5In the inferior situation, finding has again 30 blocks to lose efficacy; When test procedure continues to carry out to 3 * 10 5In the inferior situation, finding has again 30 blocks to lose efficacy; When test procedure continues to carry out to 5 * 10 5In the inferior situation, finding has again 50 blocks to lose efficacy.Can count accordingly, be 1 * 10 at the test procedure number of times 5When inferior, the number of blocks that accumulative total occurs to lose efficacy in the test of this corresponding number of times is 10; Be 2 * 10 at the test procedure number of times 5When inferior, the number of blocks that accumulative total occurs to lose efficacy in the test of this corresponding number of times is 40; Be 3 * 10 at the test procedure number of times 5When inferior, the number of blocks that accumulative total occurs to lose efficacy in the test of this corresponding number of times is 70; Be 5 * 10 at the test procedure number of times 5When inferior, the number of blocks that accumulative total occurs to lose efficacy in the test of this corresponding number of times is 120.
According to above-mentioned data, adopt the method for computer fitting to draw test curve.During curve plotting, think and satisfy the Weibull distribution relation between the number of blocks that accumulative total occurs to lose efficacy in the test of carrying out test procedure number of times and this number of times or lognormal distribution concerns.The schematic diagram of the curve shape that usually is depicted as shown in Figure 2, the horizontal ordinate of curve is the number of times of carrying out test procedure, ordinate is that the number of blocks that lost efficacy occurs accumulative total under this testing time.For clearer and more definite expression result's implication, ordinate also can be number of blocks shared ratio in all selected block number N that inefficacy occurs accumulative total in the test of this number of times.
Step S13 according to the parameter that adopts in the test, determines to adopt the equivalence inefficacy number of blocks criterion n ' of the method.
Because the method has only selected a part of block to test, it is unscientific therefore still selecting former testing standard to judge, needs to calculate the inefficacy number of blocks criterion of equivalence.The method of calculating comprises which kind of statistical models is the actual conditions that consider test determine to set up, comprise the relevance of different blocks in the rule, nonvolatile memory of the position considering block and choose, distribution and the actual situation of carrying out testing time etc., to making correction owing to the variation of choosing sample causes the situation of change of criterion in the test.
For the equivalence inefficacy number of blocks criterion n ' that sets, a kind of preferred technical scheme is to think to have in K * N the block that is selected n ' block that the probability of inefficacy occurs, should with all K * N mIn the individual block n is arranged mThe probability that individual block occurs to lose efficacy equates.The basic thought of this algorithm comprises thinks that between the block of nonvolatile memory be identical, and for each block, this statistics event that whether lost efficacy is separate and is not mutually related.Because nonvolatile memory adopts semiconductor planar technique to make usually, the process for making of each block is identical with conditionally complete, and be in each other the state of mutual electric isolation, therefore do above-mentioned hypothesis and tally with the actual situation, and this hypothesis can be simplified the computation process of equivalent inefficacy number of blocks criterion n '.Above-mentioned theory also is to calculate the statistics foundation of n '.
When equivalence inefficacy number of blocks criterion n ' is calculated, should at first analyze the probability that is in diverse location block generation inefficacy in the nonvolatile memory and how to distribute.In this embodiment, think that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mIdentical, similar with above-mentioned situation, because the process for making of each block is identical with conditionally complete, therefore does above-mentioned hypothesis and tally with the actual situation.Because in the case, the number of blocks that generation in the flash memory was lost efficacy satisfies the binomial distribution in the statistics, and the method for determining equivalent inefficacy number criterion n ' is specially to be found the solution following equation:
Σ i = 0 [ n m / K ] binomial ( i ; N m , P m ) = Σ i = 0 [ n ′ / K ] binomial ( i ; N , P m )
Binomial (i in the equation; N m, P m) implication refer to for N mThe probability that individual generation was lost efficacy is P mBlock in, have i block that the probability that lost efficacy occurs.In above-mentioned equation, N m, n m, N all is known parameters, can obtain accordingly the value of n '.
Step S14 reads in test curve with standard and carries out time numerical value T mCorresponding inefficacy number n.Find standard to carry out number of times T at curve mCorresponding point, this puts corresponding ordinate is exactly corresponding inefficacy number n.
Step S15 compares the inefficacy number n that obtains with equivalent inefficacy number of blocks criterion n ', if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
Above-mentioned is the embodiment for the reliability of nonvolatile memories method of testing provided by the invention.
Provide the embodiment for the reliability of nonvolatile memories proving installation provided by the invention below in conjunction with accompanying drawing.
Accompanying drawing 3 is depicted as the structural representation for reliability of nonvolatile memories proving installation embodiment provided by the invention.
Describedly comprise such as lower unit for the reliability of nonvolatile memories proving installation:
Block is chosen unit 200, is used for having respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block;
Loop test unit 210, be used for that block is chosen K * N the block that unit 200 is selected and carry out respectively test procedure T time, obtain test data, described test data comprises the number of inefficacy block and the number of times of each inefficacy block actual execution test procedure when losing efficacy, described T * N＜T m* N m, described T mFor standard is carried out number of times;
The test data that provides according to loop test unit 210 is provided in Drawing of Curve unit 220, and the data relationship that the number of times that statistics is carried out test procedure and the test of carrying out this number of times add up to occur between the number of blocks of inefficacy is afterwards also drawn test curve;
Criterion computing unit 230 is used for installing the parameter that adopts in test according to this, determines equivalent inefficacy number of blocks criterion n ';
Unit 240 is got in the failure number eye reading, and the test curve that is used for producing from Drawing of Curve unit 220 is read and carried out time numerical value T mCorresponding inefficacy number n;
Comparison judgment unit 250, being used for that the inefficacy number n that unit 240 obtains is got in this failure number eye reading compares with criterion computing unit 230 determined equivalent inefficacy number of blocks criterion n ', if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
Described standard is carried out number of times T mMeaning be, at all K * N mIndividual block is carried out respectively T mBehind the inferior test procedure, if the number of blocks that occurs to lose efficacy is less than inefficacy number of blocks criterion n mThen be qualified, otherwise be defective.
Described nonvolatile memory is flash memory, also can be other storeies with Long-term Preservation data characteristics.
The number of times T of described execution test procedure is greater than standard testing number of times T m
In order significantly to save the test duration, preferred testing scheme is that described N is not more than N mOne of percentage, T is not less than T m5 times.
In the described criterion computing unit, according to all K * N mIn the individual block n is arranged mThe probability of individual block inefficacy should with the K * N that an is selected sample in have the probability of the individual block inefficacy of n ' to equate principle, determine equivalent inefficacy number of blocks criterion n '.Calculate equivalent inefficacy number criterion n ' time, think that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mIdentical.The method of determining equivalent inefficacy number criterion n ' is specially to be found the solution following equation:
Σ i = 0 [ n m / K ] binomial ( i ; N m , P m ) = Σ i = 0 [ n ′ / K ] binomial ( i ; N , P m )
In the described Drawing of Curve unit, satisfy the Weibull distribution relation between the number of blocks that had lost efficacy under test procedure number of times and the corresponding number of times or lognormal distribution concerns curve plotting according to carrying out.
The scope of the value of described K is 50～100, preferred 77.
Block is chosen unit 200 and is used for selected block, the address information of selected block is committed to loop test unit 210 tests, Drawing of Curve unit 220 be used for analysis cycle test cell 210 test the result and draw test curve, and the test curve that obtains is inputted the failure number eye reading get unit 240, get unit 240 by the failure number eye reading and read this time inefficacy number n of test, choose the equivalence inefficacy number of blocks criterion n ' that the test parameter that unit 200 and loop test unit 210 adopts at work calculates with criterion computing unit 230 according to block and input together comparison judgment unit 250, obtain test result.
About the detailed description of above unit, can be with reference to content corresponding to the embodiment that is used for the reliability of nonvolatile memories method of testing.
The above only is preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (14)

1. the method for testing of a reliability of nonvolatile memories is characterized in that, comprises the steps:
Has respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block;
K * N the block that is selected carried out respectively test procedure T time, obtain test data, described test data comprises the number of inefficacy block and each inefficacy block has been carried out test procedure when occuring to lose efficacy number of times, described T * N＜T m* N m, described T mFor standard is carried out number of times;
According to test data, data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy afterwards, and adopt the method for computer fitting to draw test curve, during curve plotting, think and satisfy the Weibull distribution relation between the number of blocks that accumulative total occurs to lose efficacy in the test of carrying out test procedure number of times and this number of times or lognormal distribution concerns;
According to the parameter that adopts in the test, determine to adopt the equivalence inefficacy number of blocks criterion of the method
N ' determines that the method for equivalent inefficacy number criterion n ' is specially equation
Figure FSB00000945424800011
Find the solution binomial (i in the equation; N m, P m) implication refer to for N mThe probability that individual generation was lost efficacy is P mBlock in, have i block that the probability that lost efficacy occurs;
In test curve, read with standard and carry out time numerical value T mCorresponding inefficacy number n;
The inefficacy number n that obtains is compared with equivalent inefficacy number of blocks criterion n ', if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
2. the method for testing of described reliability of nonvolatile memories according to claim 1 is characterized in that,
Described T＞T m
3. the method for testing of described reliability of nonvolatile memories according to claim 1 is characterized in that,
Described N is not more than N mOne of percentage, T is not less than T m5 times.
4. the method for testing of described reliability of nonvolatile memories according to claim 1 is characterized in that, according to all K * N mIn the individual block n is arranged mThe probability of individual block inefficacy should with the K * N that an is selected sample in have the probability of the individual block inefficacy of n ' to equate principle, determine equivalent inefficacy number of blocks criterion n '.
5. the method for testing of described reliability of nonvolatile memories according to claim 4 is characterized in that, calculates equivalent inefficacy number criterion n ' time, thinks that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mIdentical.
6. the method for testing of described reliability of nonvolatile memories according to claim 1 is characterized in that, described nonvolatile memory is flash memory.
7. the method for testing of described reliability of nonvolatile memories according to claim 1 is characterized in that, the scope of the value of described K is 50～100.
8. the proving installation of a reliability of nonvolatile memories is characterized in that, described device comprises:
Block is chosen the unit, is used for having respectively N at K mChoose respectively N block in each of the nonvolatile memory of individual block;
The loop test unit, be used for that block is chosen K * N the block that the unit is selected and carry out respectively test procedure T time, obtain test data, described test data comprises the number of inefficacy block and each inefficacy block has been carried out test procedure when occuring to lose efficacy number of times, described T * N＜T m* N m, described T mFor standard is carried out number of times;
The Drawing of Curve unit, be used for according to test data, data relationship between the number of blocks that the number of times of statistics execution test procedure and the test of this number of times of execution add up to occur to lose efficacy afterwards, and adopt the method for computer fitting to draw test curve, during curve plotting, think and satisfy the Weibull distribution relation between the number of blocks that accumulative total occurs to lose efficacy in the test of carrying out test procedure number of times and this number of times or lognormal distribution concerns;
The criterion computing unit is used for installing the parameter that adopts in test according to this, determines equivalent inefficacy number of blocks criterion n ', determines that the method for equivalent inefficacy number criterion n ' is specially equation
Figure FSB00000945424800021
Find the solution binomial (i in the equation; N m, P m) implication refer to for N mThe probability that individual generation was lost efficacy is P mBlock in, have i block that the probability that lost efficacy occurs;
The unit is got in the failure number eye reading, and the test curve that is used for producing from the Drawing of Curve unit is read with standard and carried out time numerical value T mCorresponding inefficacy number n;
Comparison judgment unit, being used for that the inefficacy number n that the unit obtains is got in this failure number eye reading compares with the determined equivalent inefficacy number of blocks criterion n ' of criterion computing unit, if n＜n ', then this batch nonvolatile memory is qualified, if n 〉=n ' then is defective.
9. the proving installation of described reliability of nonvolatile memories according to claim 8 is characterized in that T＞T m
10. the proving installation of described reliability of nonvolatile memories according to claim 8 is characterized in that, described N is not more than N mOne of percentage, T is not less than T m5 times.
11. the proving installation of described reliability of nonvolatile memories is characterized in that according to claim 8, in the described criterion computing unit, according to all K * N mIn the individual block n is arranged mThe probability of individual block inefficacy should with the K * N that an is selected sample in have the probability of the individual block inefficacy of n ' to equate principle, determine equivalent inefficacy number of blocks criterion n '.
12. the proving installation of described reliability of nonvolatile memories according to claim 11, it is characterized in that, in the described criterion computing unit, calculate equivalent inefficacy number criterion n ' time, think that the probability P that lost efficacy occurs to be in the diverse location block in the nonvolatile memory mIdentical.
13. the proving installation of described reliability of nonvolatile memories is characterized in that according to claim 8, described nonvolatile memory is flash memory.
14. the proving installation of described reliability of nonvolatile memories is characterized in that according to claim 8, the scope of the value of described K is 50～100.
CN 200810040767 2008-07-18 2008-07-18 Method and device for testing reliability of nonvolatile memories Active CN101630534B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200810040767 CN101630534B (en) 2008-07-18 2008-07-18 Method and device for testing reliability of nonvolatile memories

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200810040767 CN101630534B (en) 2008-07-18 2008-07-18 Method and device for testing reliability of nonvolatile memories

Publications (2)

Publication Number Publication Date
CN101630534A CN101630534A (en) 2010-01-20
CN101630534B true CN101630534B (en) 2013-03-27

Family

ID=41575605

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200810040767 Active CN101630534B (en) 2008-07-18 2008-07-18 Method and device for testing reliability of nonvolatile memories

Country Status (1)

Country Link
CN (1) CN101630534B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103489485A (en) * 2013-09-04 2014-01-01 深圳市江波龙电子有限公司 Flash memory quality detection method and device
CN103700408B (en) * 2014-01-07 2017-03-29 上海华虹宏力半导体制造有限公司 The detection method of memorizer
CN105097047B (en) * 2014-05-04 2017-12-29 中芯国际集成电路制造(上海)有限公司 Memory, the detection circuit of storage array and method
CN104347121B (en) * 2014-10-28 2018-05-01 武汉新芯集成电路制造有限公司 A kind of filler test method of reliability of flash memory
CN106951701B (en) * 2017-03-15 2019-03-05 珠海全志科技股份有限公司 The calculation method and system of NAND FLASH service life
CN108831517B (en) * 2018-05-23 2021-04-27 武汉忆数存储技术有限公司 Method and test device for judging reliability of flash memory chip based on operation time or current
CN109448774B (en) * 2018-10-15 2021-03-19 上海华虹宏力半导体制造有限公司 Method for judging interference storage area position of flash memory
CN110428863A (en) * 2019-07-24 2019-11-08 深圳忆联信息系统有限公司 A kind of sampling observation method and device thereof of flash memory particle long-term reliability
CN112003762A (en) * 2020-08-07 2020-11-27 苏州浪潮智能科技有限公司 Method and device for testing high reliability of storage device

Also Published As

Publication number Publication date
CN101630534A (en) 2010-01-20

Similar Documents

Publication Publication Date Title
CN101630534B (en) Method and device for testing reliability of nonvolatile memories
CN102651054B (en) Probability method of electronic product service life model based on Bayesian theory
US20160320451A1 (en) Simulation verification method for fpga function modules and system thereof
Jeong et al. A fast built-in redundancy analysis for memories with optimal repair rate using a line-based search tree
CN103383659B (en) A kind of software acceleration lifetest method
Fan et al. Improved volume diagnosis throughput using dynamic design partitioning
CN103942138A (en) Testing method and device
CN103150228A (en) Synthesizable pseudorandom verification method and device for high-speed buffer memory
CN105469834A (en) Testing method for embedded flash memory
CN105279092A (en) Software testing method and apparatus
CN115662489A (en) Hard disk test method and device, electronic equipment and storage medium
US8848471B2 (en) Method for optimizing refresh rate for DRAM
CN111444107A (en) Software evaluation method and device
WO2019019429A1 (en) Anomaly detection method, device and apparatus for virtual machine, and storage medium
CN110473585B (en) Replacement method, device and equipment for erasing failed storage unit and storage medium
CN103065687B (en) The method of RAM production defect in parallel detection integrated circuit
CN113064828A (en) Automatic testing method and device, computer equipment and storage medium
CN104993588A (en) State monitoring alarming threshold rapid setup method
CN106971757A (en) A kind of method and system of inspection Nand Flash mass
CN112463636A (en) Test case sequencing management method and system
CN114610645B (en) Task reliability and testability joint determination method and device and computer equipment
CN103632725A (en) Erasing method and device for flash memory
US20130100752A1 (en) Method of restoring reconstructed memory spaces
CN103489485A (en) Flash memory quality detection method and device
US8301431B2 (en) Apparatus and method for accelerating simulations and designing integrated circuits and other systems

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
ASS Succession or assignment of patent right

Owner name: SEMICONDUCTOR MANUFACTURING INTERNATIONAL (BEIJING

Effective date: 20121116

C41 Transfer of patent application or patent right or utility model
TA01 Transfer of patent application right

Effective date of registration: 20121116

Address after: 201210 18 Zhangjiang Road, Shanghai, Pudong New Area

Applicant after: Semiconductor Manufacturing International (Shanghai) Corporation

Applicant after: Semiconductor Manufacturing International (Beijing) Corporation

Address before: 201210 18 Zhangjiang Road, Shanghai, Pudong New Area

Applicant before: Semiconductor Manufacturing International (Shanghai) Corporation

C14 Grant of patent or utility model
GR01 Patent grant