JPH04167537A - Test chip - Google Patents

Test chip

Info

Publication number
JPH04167537A
JPH04167537A JP29576690A JP29576690A JPH04167537A JP H04167537 A JPH04167537 A JP H04167537A JP 29576690 A JP29576690 A JP 29576690A JP 29576690 A JP29576690 A JP 29576690A JP H04167537 A JPH04167537 A JP H04167537A
Authority
JP
Japan
Prior art keywords
yield
test chip
block
blocks
lsi
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.)
Pending
Application number
JP29576690A
Other languages
Japanese (ja)
Inventor
Kiyoyuki Morita
清之 森田
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.)
Panasonic Holdings Corp
Original Assignee
Matsushita Electric Industrial Co Ltd
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 Matsushita Electric Industrial Co Ltd filed Critical Matsushita Electric Industrial Co Ltd
Priority to JP29576690A priority Critical patent/JPH04167537A/en
Publication of JPH04167537A publication Critical patent/JPH04167537A/en
Pending legal-status Critical Current

Links

Abstract

PURPOSE:To enable the yield of LSI to be easily presumed with high precision by a method wherein the title test chip is provided with a circuit block designed under specified rules and another circuit block performing the same functions having exceeding double area under the same rules as that of the former block so that the yields of respective blocks may be mutually and individually evaluated. CONSTITUTION:The title test chip is provided with a SRAM block 1 and another SRAM block 2 while respective blocks 1 and 2 are independently provided with respective pads such as power supply, address, signal line, etc., so that respective blocks 1, 2 may be composed not to be affected by the mutual defects thereby enabling the yields of respective blocks 1, 2 to be independently evaluated. The test chip is also provided with a process control monitor 3 for the process evaluation. At this time, assuming the areas of SRAM blocks 1, 2 respectively to be A3, A4, A3:A4 1:2 or more, e.g. 1:4. Through these procedures, the yield of LSI can be easily estimated with high precision.

Description

【発明の詳細な説明】 産業上の利用分野 本発明は歩留推定を行うためのテストチップに関するも
のであり、特に大規模LSI歩留を高精度で簡便に推定
するテストチップに関すム従来の技術 従来の歩留推定はチップ面積のみに着目し 歩留がチッ
プ面積のみの関数と考えていた よって、歩留推定のた
めのテストチップも単一ブロックのみで構成されたもの
を用いてい九 欠陥のポアソン分布仮定により、デバイ
ス面積をA、歩留をYとすると、 Y =EXP(−D A ’)      (1)と表
されム ここ7!、Dは適用する半導体プロセス固有の
欠陥密度であも テストチップ面積A−1歩留Y−を(
1)式に代入LDを求めも 一方(1)弐両辺の対数を
取り、 I n(Y)=−DA     (2)とし 求めたD
を(2)式に代入して歩留Yをデバイス面積Aの関数と
して表す。第3図に一例を示す。
DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a test chip for estimating yield, and particularly to a test chip for easily estimating large-scale LSI yield with high accuracy. Technology Conventional yield estimation focused only on the chip area and considered the yield to be a function of only the chip area.Therefore, test chips for yield estimation were also used that consisted of only a single block. Based on the Poisson distribution assumption, if the device area is A and the yield is Y, it is expressed as Y = EXP(-D A') (1). , D is the defect density specific to the applied semiconductor process. Test chip area A-1 yield Y- is (
Substituting LD into equation 1) On the other hand, take the logarithm of both sides of (1) and set In(Y)=-DA (2) Find D
is substituted into equation (2) to express the yield Y as a function of the device area A. An example is shown in FIG.

第3図より、面積A1を有するLSIの歩留はYlと推
定されも 発明が解決しようとする課題 しかしながら前記のような方法では 正確なLSI歩留
の推定は困難であa 例えば第4図において、第3図の
結果を用いると面積A+を有するLSIの歩留はYlと
推定されも ところな 実測値はこれより高いYlとな
り、歩留の誤差Y*−Y+を生じも この誤差Ya−Y
+4;L  実際の欠陥がランダムな分布でないために
生じも この誤差を縮小させるた&  Ferris−
Prabhuは欠陥のクラスタモデルを用いた新しい欠
陥分布を提唱した(IEEE Trans、  Sem
1cond、  Manufact、アイ・イー・イー
・イー  トランサ゛クシ1ン*ミコ7’l’99  
 v:z7yりft、Vo13.No、2.P54−5
9)。  テス  トチツブ面積をA・、歩留をY−と
L  LSI面積をA1、推定歩留をYlとすると、 と表されも ここで、bは欠陥のクラスタ率であも 全
ての欠陥が1点に集中した場合b=1、全ての欠陥がラ
ンダムに分布した場合b=oとなaよって、bのとれる
範囲は0≦b≦1であム(3)式を用いてLSI歩留を
推定するためには 適用する半導体プロセス固有の欠陥
クラスタ率すを求める必要かあ、4bを求めるためには
同一チップ中に面積の異なる複数のテストブロックが必
要である力t 従来のテストチップは単一ブロックのみ
で構成されており、 bを求めることはできないという
課題を有してい九 本発明はかかる点に鑑へ 大規模LSI歩留を高精度で
簡便に推定するためのテストチ・ンプを提供することを
目的とすも 課題を解決するための手段 本発明ζよ 所定のデザインルールで設計された回路ブ
ロックと、前記回路ブロックと同ルールで設計され 2
倍以上の面積を有する同機能の回路ブロックとを備エ 
 各々のブロックの歩留を互いに独立に評価することを
特徴とするテストチ・ツブであム 作用 本発明は前記構成により、面積の異なる2つの回路ブロ
ック(面積は各A3、A、)の歩留Ys、Y4が同時に
得られも (3)式にA3、A4、Ys、Yaを代入し
bを算出す、5bの値を再び(3)式に代入し 面積A
1のLSI歩留Y1を算出すム これにより、大規模L
SI歩留を高精度で簡便に推定することができも 実施例 第1図は本発明の実施例におけるテストチップの構成図
を示すものであ4  ′gJ1図において、テストチッ
プ中には32K bitS RA Mブロック1と12
8K bitS RA Mブロック2とを備えも 各々
のブロックは重態 アドレ入 慣号ライン等の各パッド
を独立に備えており、お互いに不良が他ブロックに影響
を及ぼさない構成になっていも 即板各々のブロックの
歩留を互いに独立に評価可能であも テストチップ内に
はプロセス評価のためのプロセスコントロールモニタ3
も備えてい432KbitSRAMブロック1の面積を
A3.128KbitSRAMブロック2の面積をA4
とすると、A s : Aa+l;4であも 第1図の
テストチップを試作すると、32KbitSRAMブロ
ック1の歩留Y2と128KbitSRAMブロック2
の歩留Y、が同時に得られも これを(3)式に代入し
て、 Ya= [Ys](A・/A・)1−bとなり、(4)
式より欠陥のクラスタ率すを算出すaあるプロセスを用
いて第1図のテストチップを試作した結果 b=0.5
となり71.ob=0.5を(3)式に代入し デバイ
ス面積と歩留の関係を求めた結果を第2図に示す。(3
)式のA1及びY・<、t  At及びYsを用い九 
本発明による歩留推定曲線11によれは 面積AS(面
積As:As=6:1)のLSI歩留はYSと推定され
4 −X  第1図のテストチップと同プロセスで試作
したLSI(面積A * : A s = 6.1)の
実測歩留はYLllとなりへ 従来法によりA婁及びY
Sを(1)式に代入してDを求め(1)式に再代入した
結果を、第2図破線で示す。従来法による歩留推定曲線
12によれば 面積AsのLSI歩留はY6’と推定さ
れ;5  YLll、Ys、Ys’を比較すると、Ys
はYI’よりもYLllに非常に近い値を示し 本発明
によるテストマスクを用いた歩留推定法が従来法よりも
精度が高いことが明かであaこのこと(よ 従来のテス
トチップは単一のブロックのみで構成されていたた教 
異なる面積を有するテストブロックを別々に試作する事
となり、半導体プロセス固有の欠陥クラスタ率すを求め
ることができなかったが 本発明のテストチップは異な
る面積の複数テストブロックを含むた敢 異なる面積を
有するテストブロックを一度に試作することができ、欠
陥クラスタ率すを求めることができも 以上のようにこの実施例によれば テストチップ中に3
2KbitS RAMブロック1と128K bitS
 RAMブロック2とを備えることにより、大規模LS
I歩留を高精度で簡便に推定することができムな耘 こ
の実施例においてテストブロック1及び2はSRAMと
した戟 DRAMとしてもよJ。
From FIG. 3, it is estimated that the yield of an LSI having an area A1 is Yl, but the problem to be solved by the invention is, however, it is difficult to accurately estimate the LSI yield using the method described above.A For example, in FIG. , using the results in Figure 3, the yield of an LSI with area A+ is estimated to be Yl. However, the actual measured value is higher than this, Yl, which causes a yield error Y*-Y+, but this error Ya-Y
+4;L This error may be caused by the fact that the actual defects are not randomly distributed. Ferris-
Prabhu proposed a new defect distribution using a defect cluster model (IEEE Trans, Sem
1cond, Manufact, IEE Transmission 1in * Miko 7'l'99
v:z7yrift, Vo13. No, 2. P54-5
9). If the test chip area is A, the yield is Y- and L, the LSI area is A1, and the estimated yield is Yl, then it can be expressed as where b is the cluster rate of defects, and all defects are reduced to one point. If all defects are concentrated, b = 1; if all defects are randomly distributed, b = o. Therefore, the range of b is 0≦b≦1, and the LSI yield is estimated using equation (3). In order to do this, it is necessary to find the defect cluster rate specific to the applied semiconductor process.In order to find 4b, it is necessary to have multiple test blocks with different areas on the same chip.The conventional test chip uses a single block. In view of this problem, the present invention provides a test chip for easily estimating the yield of a large-scale LSI with high accuracy. According to the present invention ζ, a circuit block designed according to a predetermined design rule, and a circuit block designed according to the same rule as the circuit block 2
Equipped with a circuit block with the same function that has more than twice the area.
The present invention is characterized in that the yield of each block is evaluated independently of each other.The present invention has the above-mentioned configuration to improve the yield of two circuit blocks having different areas (areas of A3 and A3, respectively). Even if Ys and Y4 are obtained at the same time, calculate b by substituting A3, A4, Ys, and Ya into equation (3).Substituting the value of 5b into equation (3) again, area A
Calculate the LSI yield Y1 of 1. As a result, large-scale LSI
SI yield can be easily estimated with high accuracy. Embodiment Figure 1 shows a configuration diagram of a test chip in an embodiment of the present invention. RAM blocks 1 and 12
Equipped with 8K bitS RAM M block 2, each block is equipped with independent pads such as critical address input, inertia line, etc., and even if the configuration is such that a defect in each block does not affect other blocks, each block can be used immediately. Although it is possible to evaluate the yield of blocks independently of each other, there is a process control monitor 3 inside the test chip for process evaluation.
The area of 432Kbit SRAM block 1 is A3. The area of 128Kbit SRAM block 2 is A4.
Then, A s : Aa + l; 4. When the test chip shown in Fig. 1 is prototyped, the yield Y2 of 32Kbit SRAM block 1 and the yield Y2 of 128Kbit SRAM block 2 are
Even if the yield Y of is obtained at the same time, by substituting this into equation (3), we get Ya = [Ys] (A・/A・)1−b, and (4)
Calculate the clustering rate of defects from the formula a Result of prototyping the test chip shown in Figure 1 using a certain process b = 0.5
Next 71. FIG. 2 shows the results of substituting ob=0.5 into equation (3) to determine the relationship between device area and yield. (3
) using A1 and Y・<, t At and Ys of the formula 9
According to the yield estimation curve 11 according to the present invention, the yield of an LSI with an area AS (area As:As=6:1) is estimated to be YS. The actual measured yield of A*: A s = 6.1) is YLll. By the conventional method, A and Y
The broken line in FIG. 2 shows the result of substituting S into equation (1) to find D and substituting it back into equation (1). According to the yield estimation curve 12 according to the conventional method, the LSI yield of area As is estimated to be Y6'; 5 Comparing YLll, Ys, and Ys', Ys
shows a value much closer to YLll than YI', and it is clear that the yield estimation method using the test mask according to the present invention has higher accuracy than the conventional method. The school was made up of blocks only.
Since test blocks with different areas were manufactured separately, it was not possible to determine the defect cluster rate specific to the semiconductor process. However, the test chip of the present invention contains multiple test blocks with different areas. As described above, according to this embodiment, the test block can be prototyped at one time, and the defect cluster rate can be determined.
2KbitS RAM block 1 and 128KbitS
By providing RAM block 2, large-scale LS
In this embodiment, test blocks 1 and 2 are SRAM, or DRAM.

SRAM、DRAM等のメモリを用いるのは不良箇所の
特定が容易であるためであるカミ 不良箇所の特定を必
要としない場合ζ上 加算器等他の機能ブロックを用い
ても良t〜 また この実施例においてメモリ容量の比
が1:4のブロック1.2を用いた力<% 1:2やl
:3のブロックを用いても良(〜 さらく この実施例
においては2個のテストブロックを用いた力丈 テスト
チップ中に3個以上のテストブロックを搭載しても良(
−テストブロック数が多いほど歩留推定精度が向上する
ことは言うまでもなu℃ 発明の詳細 な説明したよう艮 本発明によれば 大規模LSI歩留
を高精度で簡便に推定することかで叡その実用的効果は
大き1.%
Memory such as SRAM and DRAM is used because it is easy to identify the defective location.If it is not necessary to identify the defective location, other functional blocks such as adders may be used. In the example, using block 1.2 with a memory capacity ratio of 1:4, the power <% 1:2 or l
:3 blocks may be used (~ In this example, two test blocks are used.) Three or more test blocks may be mounted on the test chip (
- It goes without saying that the accuracy of yield estimation improves as the number of test blocks increases. Great practical effect 1. %

【図面の簡単な説明】[Brief explanation of the drawing]

第1図は本発明の実施例におけるテストチップの構成医
 第2図は同実施例の歩留推定結果を示す医 第3図及
び第4図は従来の方法による歩留推定結果と実測値の違
いを示す図であも1・・・32KbitS RA肱 2
・・・128KbitS RA肱 3・・・プロセスコ
ントロールモニ久 11・・・本発明による歩留推定曲
線 12・・・従来法による歩留推定曲線 代理人の氏名 弁理士 小鍜治 明 ほか2名第1図 富2図 a    AJ       A4     Asテ′
〕\゛イ ス 面→嘴A 第3図 ^o           Al デバイス興fllA I4図 ^Q           At ヂ2(イズ耐flA
Figure 1 shows the configuration of a test chip in an embodiment of the present invention. Figure 2 shows the yield estimation results of the same embodiment. Figures 3 and 4 show the results of yield estimation by the conventional method and actual measurements. The diagram showing the difference is 1...32KbitS RA arm 2
...128KbitS RA 3...Process control monitor 11...Yield estimation curve according to the present invention 12...Yield estimation curve according to the conventional method Name of agent Patent attorney Akira Okaji and two others 1st Zutomi 2 figure a AJ A4 Aste'
〕\゛Ice surface→Beak A Fig. 3^o Al device resistance flA I4Fig.^Q At ヂ2

Claims (3)

【特許請求の範囲】[Claims] (1)所定のデザインルールで設計された回路ブロック
と、前記回路ブロックと同ルールで設計され2倍以上の
面積を有する同機能の回路ブロックとを備え、各々のブ
ロックの歩留を互いに独立に評価することを特徴とする
テストチップ。
(1) A circuit block designed according to predetermined design rules and a circuit block with the same function designed according to the same rules as the circuit block and having more than twice the area, and the yield of each block can be controlled independently of each other. A test chip characterized by evaluating.
(2)請求項1に記載の回路ブロックとしてSRAMを
用いることを特徴とするテストチップ。
(2) A test chip characterized in that an SRAM is used as the circuit block according to claim 1.
(3)請求項1に記載の回路ブロックとしてDRAMを
用いることを特徴とするテストチップ。
(3) A test chip characterized in that a DRAM is used as the circuit block according to claim 1.
JP29576690A 1990-10-31 1990-10-31 Test chip Pending JPH04167537A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP29576690A JPH04167537A (en) 1990-10-31 1990-10-31 Test chip

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP29576690A JPH04167537A (en) 1990-10-31 1990-10-31 Test chip

Publications (1)

Publication Number Publication Date
JPH04167537A true JPH04167537A (en) 1992-06-15

Family

ID=17824892

Family Applications (1)

Application Number Title Priority Date Filing Date
JP29576690A Pending JPH04167537A (en) 1990-10-31 1990-10-31 Test chip

Country Status (1)

Country Link
JP (1) JPH04167537A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0872883A2 (en) * 1997-04-18 1998-10-21 Matsushita Electric Industrial Co., Ltd. Method for estimating yield of integrated circuit device
WO2006127409A2 (en) * 2005-05-20 2006-11-30 Cadence Design Systems, Inc. System and method for statistical design rule checking
JP2007201497A (en) * 1999-11-18 2007-08-09 Pdf Solutions Inc System and method for product yield prediction

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0872883A2 (en) * 1997-04-18 1998-10-21 Matsushita Electric Industrial Co., Ltd. Method for estimating yield of integrated circuit device
EP0872883A3 (en) * 1997-04-18 1998-12-16 Matsushita Electric Industrial Co., Ltd. Method for estimating yield of integrated circuit device
US6311139B1 (en) 1997-04-18 2001-10-30 Matsushita Electric Industrial Co., Ltd. Method for estimating yield of integrated circuit device
JP2007201497A (en) * 1999-11-18 2007-08-09 Pdf Solutions Inc System and method for product yield prediction
WO2006127409A2 (en) * 2005-05-20 2006-11-30 Cadence Design Systems, Inc. System and method for statistical design rule checking
WO2006127409A3 (en) * 2005-05-20 2007-02-08 Cadence Design Systems Inc System and method for statistical design rule checking
US8103982B2 (en) 2005-05-20 2012-01-24 Cadence Design Systems, Inc. System and method for statistical design rule checking

Similar Documents

Publication Publication Date Title
Mahajan et al. Embedded multidie interconnect bridge—A localized, high-density multichip packaging interconnect
Stapper Fact and fiction in yield modeling
JP2003514475A (en) System and method for product yield prediction using a logical characterization vehicle
Sato et al. A new defect distribution metrology with a consistent discrete exponential formula and its applications
US7399648B2 (en) Methods and apparatus for determining location-based on-chip variation factor
Ferri et al. Strategies for improving the parametric yield and profits of 3D ICs
JPH04167537A (en) Test chip
Stapper Small-area fault clusters and fault tolerance in VLSI circuits
US7594210B2 (en) Timing variation characterization
Li et al. FPGA as process monitor-an effective method to characterize poly gate CD variation and its impact on product performance and yield
KR19980079792A (en) Device size optimization method of semiconductor device
Sarvar et al. Effective modeling of the reflow soldering process: use of a modeling tool for product and process design
JPH09153550A (en) Pattern forming method
US20130249591A1 (en) System and method for decreasing signal integrity noise by using varying drive strengths based on likelihood of signals becoming victims
Aitken DFM metrics for standard cells
CN113467590A (en) Many-core chip temperature reconstruction method based on correlation and artificial neural network
Parks et al. The nature of defect size distributions in semiconductor processes
Chatterjee et al. Machine learning based temperature estimation for test scheduling of 3d ICs
Kim et al. Sequential modeling of via formation in photosensitive dielectric materials for MCM-D applications
Stamenkovic et al. Modeling of integrated circuit yield loss mechanisms
JP2005109417A (en) Method for evaluating simulation circuit pattern, method for manufacturing semiconductor integrated circuit, test substrate, and group of test substrates
Levasseur et al. Application of a yield model merging critical areas and defectivity to industrial products
JP2521041B2 (en) Wiring method in integrated circuit
Brewer Promise and Pitfalls of WSI
US7672749B1 (en) Method and apparatus for hierarchical process control