JPH10284564A - Method and equipment for producing semiconductor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently discriminate whether a chip can be shipped or not without using any ink mark. SOLUTION: The method for producing semiconductor is provided with a process for checking the operation of element for each chip formed on a wafer (S3-S7), process for recording the result information for each chip into a data base (S8) and process for dividing the wafer for each chip, taking out only the satisfactory chips through a die bonder and packaging them based on the information in the data base (S11-S16). The equipment for producing semiconductor is provided with a information fetching means for fetching the result information of the operation check of element for each chip formed on the wafer and a chip selecting means for performing processing only to the satisfactory chips based on the information provided by the information fetching means.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor manufacturing method and a semiconductor manufacturing apparatus for inspecting an element for each chip formed on a wafer and packaging a non-defective product.

[0002]

2. Description of the Related Art In a conventional semiconductor manufacturing method, after performing a device inspection for measuring the operating characteristics of elements formed in a plurality of chips on a wafer in a wafer process, it is not possible to distinguish between good and defective products. Ink marks are provided on good chips.

[0003] Thereafter, when performing an appearance inspection of the device, for example, a chip with an ink mark on the wafer is determined to be defective with an optical microscope, and this inspection is not performed, and the appearance of only a good chip without the ink mark is determined. Inspection is performed. When a new defect is detected in the appearance inspection, an ink mark is attached to the chip.

[0004] Next, in the assembling process, dicing is performed in which the wafer is attached to an ultraviolet irradiation-curable adhesive tape and cut into chips. After the adhesive tape is irradiated with ultraviolet rays to reduce the adhesiveness, the appearance is inspected again. If a defect is detected here, an ink mark is attached to the chip.

[0005] The wafer cut for each chip is transported to the next die bonder while being attached to an adhesive tape. In this die bonder, the presence or absence of the ink mark on the conveyed chip is recognized and determined one by one, and the chip with the ink mark is not picked up, and only the chip without the ink mark is picked up. I have.

Here, even if the ink mark is attached,
For example, a chip in which only a part of the chip can be used (hereinafter, simply referred to as “half bit product”) is provided with an ink mark different from a defective product. The half-bit product is once determined to be defective during device inspection, but is in a state where only a part of the chip can be used by the subsequent repair processing, for example, in an unusable part of the chip, A different number of ink marks than completely defective products are applied.

[0007] For example, one ink mark is attached to a completely defective product, and two ink marks are attached to a half bit product to distinguish them.

The die bonder recognizes the presence or absence of an ink mark on the chip, and picks up a good chip without the ink mark. As a result, a defective product and a half bit product remain in the die bonder. The half-bit products are separately arranged on a pallet, and later mounted on a die bonder to package only the half-bit products.

[0009]

However, in such a method of determining whether a chip is good or defective by attaching an ink mark, variations in the amount of ink ejected in each step of applying the ink mark, ink marks, etc. The uncertain position causes a recognition failure in the die bonder.

In addition, when this ink scatters to a non-defective product and is erroneously recognized in a die bonder and is determined to be a defective product, or when the ink scattered to a non-defective product and a half-bit product ink are packaged with a mold resin, the package is formed. Cause swelling, which causes a decrease in reliability.

Further, after the ink mark is attached,
Heating in a baking oven to dry the ink has an adverse effect on the quality of the device due to the heat.

In a semiconductor manufacturing apparatus such as a visual inspection apparatus, a chip sorter, and a die bonder which performs processing by recognizing the presence or absence of an ink mark, it is necessary to sequentially detect and recognize the presence or absence of an ink mark for all chips. Yes, it takes a lot of time to determine the quality.

[0013]

SUMMARY OF THE INVENTION The present invention is directed to a semiconductor manufacturing method and a semiconductor manufacturing apparatus made to solve such a problem. That is, the semiconductor manufacturing method of the present invention includes a step of performing an operation inspection of an element for each chip formed on a wafer, a step of recording pass / fail information of each chip in the operation inspection in a database, and a step of: After the division, only a non-defective chip is taken out by a die bonder and packaged based on the pass / fail information of the database.

Further, the semiconductor manufacturing apparatus according to the present invention is characterized in that a good or bad information obtaining means for obtaining good or bad information in an operation inspection of an element for each chip formed on the wafer, and a good or bad product based on the good or bad information obtained by the good or bad information obtaining means. And a chip selecting means for performing a process only on the chip having the following condition.

According to the semiconductor manufacturing method of the present invention, after the operation of the element is inspected, the pass / fail information is recorded in the database, so that only the non-defective chips are efficiently taken out by using the database in the die bonder and packaged. Can be

Further, in the semiconductor manufacturing apparatus of the present invention, since the pass / fail information in the operation inspection of the element is obtained by the pass / fail information acquisition means, only the non-defective chips are selected by the chip selecting means based on the pass / fail information. Processing can be performed. In other words, since it is not necessary to perform processing for defective products, efficient production can be achieved.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a flowchart illustrating a semiconductor manufacturing method according to the present embodiment, and FIG. 2 is a system configuration diagram in semiconductor manufacturing.

That is, as shown in FIG. 1, the semiconductor manufacturing method of this embodiment mainly includes steps S1 to S
9 and an assembling process in steps S10 to S15 and thereafter.

First, as a step S1 of the wafer processing step, a polyimide as a protective film is formed, and after a step S2 is inputted, a TEG (Test Element G) is formed in a step S3.
roup) measurement, and then 1st shown in step S4
Probing, database record shown in step S5,
Repair shown in step S6, 2nd shown in step S7
Probing.

Then, the database recording shown in step S8 and the wafer completion inspection shown in step S9 are performed, and finally the back grinding shown in step S10 is performed.

In the assembly process, first, in step S11
The wafer is loaded as shown in step S12, the wafer is mounted in step S12, the dicing is shown in step S13, the UV (ultraviolet ray) is irradiated in step S14, the dice is sorted in step S15, and the dice bond is shown in step S16.
Then, packaging is performed through a mold sealing step (not shown).

In the present embodiment, the first probing (step S
4) Database record after (step S5) and 2n
It is characterized in that, as the database record (step S8) after the d-probing (step S7), information on the quality of the operation inspection is recorded in the database.

In the assembling step, the database of the quality information recorded in step S7 is used to execute step S1.
In addition to inspecting only the chips (dies) that have become non-defective in the dice sorting shown in FIG. 4, only non-defective chips are picked up by the die bonder shown in step S15 and mounted on a lead frame or the like.

This eliminates the necessity of performing the pass / fail information using the ink marks, and can solve the problems of ink splattering and time in selecting non-defective products. That is, the problem of ink scattering is eliminated by not using the ink mark. In addition, when ink marks are used, whether or not all the chips have the ink marks is sequentially determined by performing image processing on images obtained optically, and it takes a lot of time to determine good products. However, by using the database, only the non-defective chips can be immediately selected without performing image acquisition and image processing, so that the time required for the quality selection can be greatly reduced.

The system configuration diagram shown in FIG. 2 shows a connection configuration of various devices in an inspection process and an assembly process.
That is, as the inspection process, the host computer H that manages the devices and the like in the probing process in the wafer process.
1. A database D1 for storing pass / fail information in an operation inspection, a prober P for performing an operation inspection, an ink marker M to be used as required, an appearance inspection machine S including a microscope for inspecting the appearance of a wafer, and a network N such as a LAN. Connected through.

The assembling process includes host computers H2 and H3 for managing the assembling process, databases D2 and D3 having the same pass / fail information as the inspection process, a terminal T,
Wafer mounter WM, dicer WD, UV irradiator L, sorter CS via personal computer PC22, die bonder D via personal computer PC23
B is connected via a network N connecting the inspection processes. A barcode printer BP is connected to the wafer mounter WM via a personal computer PC21.

An optical reading device (OCR) for reading a wafer ID (identification code) is incorporated in the prober P, the appearance inspection device S, and the wafer mounter WM.

Further, the appearance inspection machine S in the present embodiment,
A semiconductor manufacturing apparatus such as a sorter CS, a die bonder DB, or the like has a pass / fail information obtaining unit that obtains pass / fail information indicating a result of an operation test of an element recorded in the databases D1, D2, and D3, and a pass / fail information based on the pass / fail information. And a chip selecting means for performing processing on only the chips. In addition,
The pass / fail information acquiring means and the chip selecting means may be those which are programmed by a personal computer connected to the semiconductor manufacturing apparatus.

FIG. 3 is a plan view for explaining the state of the wafer. That is, in the inspection process, the wafer is transferred in a state of a wafer W as shown in FIG. 3A, and the wafer ID assigned to the wafer W is read by the optical reading device (OC) described above.
R) to determine the wafer.

Further, the wafer mounter W in the assembly process
In M (see FIG. 2), as shown in FIG. 3 (b), the wafer W and the ring R are bonded with an ultraviolet irradiation curing type adhesive tape TP, and the wafer W is divided into chips by the dicer WD. Each chip can be held. Further, a barcode label BL created by a barcode printer BP (see FIG. 2) can be attached to the adhesive tape TP.

In this system configuration diagram, arrow A indicates the flow of test results at prober P, and arrow B indicates the flow of test results (A) at prober P plus visual inspection results.

Next, the semiconductor manufacturing method of the present embodiment with such a system configuration will be described in order. In the following description, FIG. 2 will be referred to unless otherwise specified. First, the wafer W to be inspected by the prober P (FIG.
(See (a))) (not shown)
Is set, and the lot No. , Probe card No. Is input, and the start key provided on the prober P is pressed.

The prober P is a wafer W in a wafer carrier.
At this time, the wafer ID attached to the wafer W is read by an optical reading device (OCR), and is notified to the host computer H1 in the inspection process via the network N.

The host computer H1 creates directories and files as shown in FIG. 4 based on the information received from the prober P. FIG. 4 shows a directory name and a file name. For example, as shown in FIG. 4A, “FWWW-XXXX” is used as the wafer ID.
-YY-ZZ ”is read by the optical reader (OCR) of the prober P, and FIG.
The directory name “WWWWXXX” as shown in FIG.
X ".

Here, the first five digits of the wafer ID "FWW
“WW” is the model code of the semiconductor device to be manufactured, the next four digits “XXXX” are the lot number, the next two digits “YY” are the wafer number, and the last two digits “ZZ” are the checksum. As the directory name, “WWWWXXXX” is given using the model code and the lot number in the wafer ID.

As shown in FIG. 4C, “XXXXXXY.DAT” is given as a file name using, for example, a lot number and a wafer number in a wafer ID.

As shown in FIG. 4D, one directory (for example, “WWWWXXXX”) is assigned one corresponding to one model code or lot number.
One file name is assigned to one wafer. Therefore, one directory contains a plurality of units belonging to the same model code or lot number (for example,
File names corresponding to 25 wafers (for example, “X
XXX01. DAT ”to“ XXXX25.DAT ”
5 files).

The measurement result for each chip measured by the prober P is passed to the host computer H1 in the inspection process as a signal (pass / fail information) indicating good or bad. The host computer H1 creates a database D1 based on the pass / fail information.

FIG. 5 is a diagram for explaining records in the database, and FIGS. 6 and 7 are diagrams for explaining the data structure in the database. As shown in FIG. 5, when a plurality of chips (see a rectangular portion in the figure) are formed in a matrix on one wafer, one row along the horizontal direction in the figure is regarded as one record as a database. , A structure including data for a plurality of records along the vertical direction in the figure.
Note that the number of data in each record corresponds to the number of chips constituting one column (for example, the number of data of record “1” = 4, the number of data of record “2” = 6), and does not always match.

In one file, data is stored for each record corresponding to one column as shown in FIG. For example, as the data of the first record number, the data of the map start chip is stored at the beginning. Here, the number of records for one wafer and the distance (XDistanc) from the center of the wafer to the map start chip (see FIG. 5)
e, YDistance) are stored.

After the data of the map start chip, the data of the first record number (the address of the chip at the head of the column and the measurement data thereafter (for example, up to 10
5) are stored.

After the data of the second record number, only the measurement data excluding the data of the map start chip in the data of the first record is stored.

Further, as shown in FIG. 7, each measurement data in one record has a 2-byte configuration, and the first one of the two bytes has a measurement history and a measurement result of "0". "1", and the subsequent one byte stores BINDATA indicating detailed information in the pass / fail information.

For example, the measurement result (PASS = 0, FAIL = 1) up to a total of three measurement results (T1 to T3) can be recorded in the first byte of the data shown in FIG. In the case of a chip that has undergone an appearance inspection, “1” is input to a bit indicated by “7” in the figure, and for an unmeasured chip, “1” is input to a bit indicated by “6” in the figure. In the case of a chip for which measurement was skipped, "5"
“1” is input to the bit indicated by “”, and “1” is input to the bit indicated by “4” in the chip for forcibly attaching the ink mark.
Is entered. As a result, the measurement history can be understood.

Binary data corresponding to a code indicating detailed information of measurement is input to BIN DATA input to one byte after the data. For example, if it is defective, binary code corresponding to a code consisting of "09H (H is a code indicating hexadecimal; the same applies hereinafter)" is input, and if it is completely good, a binary code corresponding to a code consisting of "01H" is input. The data is entered.

Further, the device is remedied in the repair process, and although it is a non-defective device, some functions can be used, and a non-defective product having no problem in quality is inputted with a code consisting of "02H" and binary data corresponding thereto. You. In this way, detailed information that is more detailed than simple pass / fail information is set as a code, and binary data corresponding to this code is
By storing in NDATA, finer sorting can be performed.

FIG. 8 is a diagram for explaining BIN DATA. The BIN DATA is set, for example, for each model, and corresponds to the model code “FWWWW” shown in FIG. 4A, and the BIN DATA file name “FWWWWW.D”.
AT "is set.

In the "BINLT" directory in the disc, BIN DATA files for each model are stored. Each BIN DATA file describes the correspondence between code data and detailed information. For example, the BIN DATA file “FWWW
W. In the "DAT", detailed information is described corresponding to nine codes "01" to "09".

Here, the code "01" is good only in the memory area of 1M (megabyte), the code "02" is good only in the memory area of 2M (megabyte),..., Code "09"
Indicates FAIL, that is, a completely defective product. Note that BIN DATA is "00" to "F
256 codes up to "F" can be registered.

The binary data corresponding to the BINDATA code is stored in the last one byte of the two bytes of the data of each chip shown in FIG. 7, and detailed information obtained by the measurement result of the chip Can be obtained in a later step.

In the prober P, the creation of the database D1 is repeated until the measurement of all the wafers in the wafer carrier is completed. The database is also stored on the floppy disk FD in consideration of the case where the host computer H1 has stopped.

Next, in the first probing with the prober P, the repair is performed by a computer which manages the determination as to whether or not a non-defective product (for example, the data of the measurement result T1 shown in FIG. 7 is "0") requires a separate repair. Repair processing is performed on the necessary chips. In this determination, an AND process of the value of the measurement result T1 shown in FIG. 7 and the value in the repair management computer is used.

Next, 2nd probing is performed. 2n
In the case of d probing, FIG.
Are performed only when the value of T1 in the data shown in FIG. As a result, the operation for the defective chip can be omitted altogether, and the required 2nd probing can be performed in a short time. The measurement result of the second probing is written to a bit at T2 of the data shown in FIG.

Next, a wafer completion inspection (appearance inspection) is performed. In this inspection, if "1" is input to either TI or T2 of the data (see FIG. 7) of the chip to be inspected, it is determined that the chip is defective, and the inspection is skipped. When both T1 and T2 of the data are “0”, it is determined that the data is non-defective and the appearance inspection is performed by the appearance inspection machine S.

When it is determined in the wafer completion inspection that the product is defective, the BIN of the data shown in FIG.
Code "09H" indicating that it is defective in DATA
And the corresponding binary data.

The measurement result is sent to the host computer H1 in the inspection process via the network N. At this time, it is also sent to the floppy disk FD at the same time, and even if the host computers H2 and H3 in the assembling process described later cannot access the host computer H1 in the inspection process,
Pass / fail information can be obtained using the floppy disk FD.

Next, the wafer having undergone the inspection process is transported to the assembly process while being housed in a wafer carrier. In addition, the floppy disk FD in which the inspection result data is stored is transferred to the assembling process together with the transfer of the wafer carrier.

In the assembling process, first, the process of charging the wafer conveyed at the terminal T into the assembling process is started. The input start operation is performed on a lot No. called an AT number. Start by typing

Host computers H2 and H3 in the assembling process
Stores information relating to the association between the AT number and the wafer ID in advance, and when the AT number is input, information on the wafers stored in the wafer carrier, that is, the inspection from the host computer H1 in the inspection process is performed. A process for copying the obtained data via the network N is performed.

Host computers H2 and H3 in the assembly process
Are created with names as shown in FIG. 9, for example. One directory is created corresponding to one AT number, and one file is created corresponding to one wafer. In the directory, files for the number of wafers stored in the wafer carrier are created. become. The data structure in the file is shown in FIGS.
Is the same as the data structure in the inspection process shown in FIG.

When the loading into the assembling process is completed, the wafer carrier is transported to the wafer mounter WM. In the wafer mounter WM, a wafer is loaded from a wafer carrier, and the wafer W is attached to a UV irradiation curing type adhesive tape TP together with a ring R as shown in FIG.

When the bonding of the adhesive tape TP is completed, the wafer ID is read by an optical reading device (OCR) incorporated in the wafer mounter WM, and the bar code printer BP connected based on the wafer ID is read.
Thus, the barcode label BL is attached to the designated position of the adhesive tape TP (see FIG. 3B).

The contents of the bar code label BL are almost the same as the contents of the wafer ID, and are, for example, bar codes corresponding to those of the wafer ID shown in FIG. .

The process of attaching the bar code label BL is for assisting in the recognition failure in the optical reading device (OCR). Can be read.

The wafer to which the barcode label BL has been attached is transported together with the floppy disk FD to the next step, the dicer WD, and then to the sorter CS via an ultraviolet irradiation step. This floppy disk FD is inserted into the sorter CS. In the sorter CS, the magazine No. of the magazine in which the wafer is stored.
(Barcode) is read by a barcode reader,
The magazine No. is sent to the host computers H2 and H3 in the assembly process. And the host computers H2 and H3 check the AT number, product name, type code, number of wafers, process code, etc., which are held as binding information.

If this check is normal, information on the AT number, product name, product code, and number of wafers is transmitted to the connected personal computer PC22. Next, the sorter CS loads one wafer from the magazine,
The bar code label BL (see FIG. 3B) is read by a bar code reader.

Then, the read ID is notified to the personal computer PC22, and the personal computer P22 is notified.
In C22, the necessary directories and files are accessed and copied from the received IDs to the host computers H2 and H3. The copied data is stored on a hard disk or the like of the personal computer PC22.

However, copying all the files under the directories stored in the databases D2 and D3 of the host computers H2 and H3 requires a lot of time. In this case, it is necessary to deal with the wafer directly related to the received ID. Copy only one file.

If there is no corresponding file in the databases D2 and D3 of the host computers H2 and H3, error processing is performed. After the wafer has been cut by the dicer WD, the sorting machine CS starts the inspection based on the pass / fail information stored in the corresponding file. The sorter CS inspects only good chips based on the value of BIN DATA (see FIG. 7) in the data.

If it is determined by this inspection that the data is defective, the BIN DATA of the data of the corresponding chip is rewritten to a code indicating the defect. BIN DATA rewrites not only the file copied to the hard disk or the like of the personal computer PC22, but also the file of the floppy disk FD inserted into the sorter CS. Also,
The sorting machine CS is provided with a counter for each defective category, and when the inspection is completed, the non-defective products, the number of defective products in the wafer, the data for each defective category, etc. are copied to the personal computer PC22 for each file.

The personal computer PC22 transmits the contents of this file to the host computers H2 and H3. The host computers H1 and H2 overwrite the received file with the file before reception. While processing the first wafer, the personal computer PC
Reference numeral 22 accesses the host computers H2 and H3 to copy files corresponding to the second and subsequent wafers. As a result, it is not necessary to stop the apparatus in the processing of the second and subsequent sheets.

Next, the wafer is conveyed to the die bonder DB, and the floppy disk FD is inserted into the personal computer PC 23 connected to the die bonder DB to start the processing. In addition, floppy disk FD
Is used only when the host computers H2 and H3 are stopped.

After setting the magazine containing the wafers in the die bonder DB, the personal computer PC2
No. 3 is the magazine No. (Barcode) is read by a barcode reader, and the corresponding magazine No. is read to the host computers H2 and H3. Notify.

The host computers H2 and H3 send the received magazine No. Then, the process code is checked, and if it is normal, OK is returned to the personal computer PC23 to set a notification flag. Next, when the process start button of the die bonder DB is pressed, a notification indicating the start of the process is sent to the host computers H2 and H3 via the personal computer PC23.

The host computers H2 and H3 send the notice and the magazine No. Based on both the notification flag and
The remote start is notified to the personal computer PC23, and the personal computer PC23 notifies the die bonder DB. Upon receiving this notification, the die bonder DB enters the operating state.

The die bonder DB loads one wafer from the set magazine, and reads the bar code label BL affixed to the adhesive tape TP (see FIG. 3B) on the alignment stage with a bar code reader. Then, the read ID is notified to the personal computer PC23, and the personal computer PC23 copies the necessary directories and files from the databases D2, D3 of the host computers H2, H3 based on the notified ID. The copied database is stored on a hard disk or the like of the personal computer PC23.

However, copying all the files under the directories stored in the databases D2 and D3 of the host computers H2 and H3 requires a lot of time. In this case, it is necessary to deal with the wafer directly related to the received ID. Copy only one file.

At this time, the personal computer PC
Reference numeral 23 also receives map information on the chip layout of the die bonder DB from the die bonder DB. If there is no corresponding file in the databases D2 and D3 of the host computers H2 and H3, error processing is performed.

In the die bonder DB, since the captured file (map information) cannot be applied as it is, the personal computer PC23 performs coordinate conversion of the map information.

That is, in the map coordinates from the inspection process to the die selection process in the assembly process, the origin is located at the upper left corner with the wafer orientation flat (hereinafter, simply referred to as “orientation flat”) on the left side of the wafer, and the die bonder DB The upper right is the origin. Further, the former has data of only the inspection chip, while the latter has data as a rectangular array based on the maximum number of X and Y axes.

Further, with respect to the angle of the orientation flat (notch), the former has the angle information in the clockwise direction with the top (arbitrary) of the wafer being 0 degrees, while the latter has the angle information in the clockwise direction with the bottom (arbitrary) of the wafer being 0 degrees. Having information, the notch position angle of the ring R (see FIG. 3B) and the relative orientation flat angle based on the notch position angle are determined.

To determine the position angle of the orientation flat (notch) in the die bonder DB, the following equation (1) is applied.

(Notch angle of ring R + relative orientation flat angle) / 360 (1)

After performing the coordinate conversion of the map information, the personal computer PC 23 sends the converted map information to the die bonder DB. The die bonder DB starts picking up chips based on the map information. At this time, referring to the value of BIN DATA of the data, only the chip in which the binary data corresponding to the code indicating the non-defective product is written is picked up. This eliminates the need to pick up defective chips.

In the above-described embodiment, in the first probing, a pass / fail judgment such as a failure when the inspection fails and a non-defective product when the inspection passes is performed. ) Is changed, for example, even if the test fails at the speed rank of 60 ns, if the test passes at the speed rank of 80 ns, the corresponding chip can be used as a non-defective product at the speed rank of 80 ns.

That is, if the measurement is failed in the first probing under the predetermined measurement conditions, the measurement with the lowered rank is performed only by changing the program without exchanging the members such as the substrate and the jig. If it passes the rank, it can be determined immediately, and if the code corresponding to BIN DATA is written at that time, it is possible to perform the rank classification at the stage of the first probing.

When a half bit product can be collected after the repair process, the BIN of the corresponding chip at that time is obtained.
If code data indicating a half-bit product is input in DATA, the BI bit will be
By referring to the code data input in NDATA, it is possible to immediately determine what kind of half-bit product the chip is.

The system configuration of the present embodiment described above is not limited to that shown in FIG. 2, but can be applied to other system configurations.

[0089]

As described above, according to the semiconductor manufacturing method and the semiconductor manufacturing apparatus of the present invention, the following effects can be obtained. In other words, since the pass / fail information using the database is applied in the inspection process, it is possible to prevent a variation in the ink ejection amount when using the ink mark and a recognition failure in the die bonder due to the indefinite position of the ink mark. It is possible to reduce the time required for determination and production.

Further, since the package does not bulge due to the ink adhering to the die bonder due to scattering of the ink to a non-defective product, the baking for drying the ink is not required, and the reliability of the semiconductor device is reduced. It can be improved.

Further, according to the semiconductor manufacturing apparatus of the present invention, since the pass / fail information in the element operation test is obtained by the pass / fail information acquisition means, only the non-defective chips are selected by the chip selecting means based on the pass / fail information. In addition, efficient production can be performed without processing defective products.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a semiconductor manufacturing method according to an embodiment.

FIG. 2 is a system configuration diagram.

FIG. 3 is a plan view illustrating a state of a wafer.

FIG. 4 is a diagram showing directory names and file names.

FIG. 5 is a diagram illustrating a record.

FIG. 6 is a diagram (part 1) for explaining a data structure;

FIG. 7 is a diagram (part 2) for explaining the data structure;

FIG. 8 is a diagram illustrating BIN DATA.

FIG. 9 is a diagram showing directory names and file names in an assembly process.

[Explanation of symbols]

D1, D2, D3 Database DB
Die bonder H1, H2, H3 Host computer P Prober S Visual inspection machine WD
Dicer WM Wafer mounter

Claims (7)

[Claims] A step of performing an operation test of an element for each chip formed on the wafer; a step of recording pass / fail information of each chip in the operation test in a database; and a step of dividing the wafer into chips. A step of extracting only non-defective chips with a die bonder based on the pass / fail information of the database and packaging the chips. 2. The semiconductor manufacturing method according to claim 1, wherein the pass / fail information of each chip in the operation test is passed to at least the die bonder via a predetermined network. 3. The semiconductor manufacturing method according to claim 1, wherein said pass / fail information includes coded detailed information on pass / fail. 4. The database according to claim 1, wherein a directory is provided for each predetermined type to which the wafer belongs, and a file for providing pass / fail information for each chip of the wafer is provided in the directory. 2. The semiconductor manufacturing method according to 1. 5. The semiconductor manufacturing method according to claim 1, wherein a barcode label corresponding to the identification code attached to the wafer is created, and the barcode label is attached and transported together with the wafer. 6. A good / bad information obtaining means for obtaining good / bad information in an operation inspection of an element for each chip formed on a wafer, and based on good / bad information obtained by said good / bad information obtaining means, only a good chip is determined. A semiconductor manufacturing apparatus comprising: a chip selecting unit for performing a process. 7. The semiconductor manufacturing apparatus according to claim 6, wherein said pass / fail information acquiring means obtains a coded version of detailed information in said pass / fail information.