JP4384899B2 - Device chip position measurement method - Google Patents

Description

  The present invention relates to a plurality of semiconductor devices (in a WLCSP (Wafer-level Chip Scale Package) process, etc.) that are separated into pieces by a dicer and arranged on a dicing tape from a wafer or strip on which a plurality of semiconductor devices are formed. Hereinafter, it is also referred to as a “device chip”).

  Handheld products such as mobile phones and digital cameras continue to be improved to meet the demand for smaller, thinner and lighter products. In order to solve these problems, it has become a problem to reduce the mounting volume and mass of LSI (large-scale integration) itself (that is, to reduce the area and thickness).

  CSP (Chip Scale Package) is often used as a method to reduce the mounting volume and mass of LSI itself. CSP is a term that refers to a semiconductor device that is packaged such that the area of the semiconductor chip and the finished product as an LSI component have substantially the same area (area of about 115% or less with respect to the area of the semiconductor chip). If the terminals are arranged around the semiconductor chip, the package size becomes large. Therefore, the CSP employs a method of arranging the terminals in a matrix on the back surface of the semiconductor chip.

  CSP is manufactured by two main assembly methods. In the first method, after dicing from a wafer with a dicer and separated into individual pieces, the picked semiconductor chips are arranged in a matrix on a rectangular lead frame, and molding, wiring, and terminal formation are performed thereon. A thing formed in this way is called a strip. The semiconductor devices formed in the strip are separated into individual semiconductor devices by a dicer.

  The second method is a method in which molding, wiring, and terminals are formed on the surface on which each semiconductor chip is formed as it is on the wafer, and the thus formed method is called WLCSP. In the WLCSP process, semiconductor devices are separated into individual semiconductor devices by a dicer.

  The outer shape of the strip is rectangular, and the outer shape of the WLCSP is a wafer shape (nearly circular), but in any case, after the packaging process such as molding is completed, it is separated into pieces by a dicer. Before shipping, it is necessary to perform a semiconductor device characteristic inspection called a final test to select the semiconductor device.

  The characteristic inspection of the semiconductor device is performed using two apparatuses. One is a device that generates an electrical signal for characteristic inspection, and is sometimes called a device chip tester. The other is an apparatus that identifies and controls a semiconductor device to be inspected, and is sometimes called a prober (CSP prober or CSP handler). In a prober, a precise positioning mechanism is required to identify each semiconductor device on the wafer. The two devices are connected from a device chip tester to a measurement electrode installed in a prober by a measurement cable. In the following description, the characteristic inspection of a semiconductor device is sometimes referred to as a test or a testing.

  There are three methods for determining the final stage of the semiconductor device manufacturing process, each of which has advantages and disadvantages.

  The first method of the final test is a method of measuring electrical characteristics of a semiconductor device before the semiconductor device is separated into pieces by a dicer. In this case, since it is before singulation, at the time of measuring the electrical characteristics of the semiconductor device, the positional relationship between the semiconductor devices is strictly maintained as designed, and the electrode position on the measuring device side and the semiconductor This is advantageous because the position of the device side terminal can be perfectly matched. On the other hand, since the process of singulation with a dicer is performed after measurement, the influence cannot be predicted. In other words, even if it is a good product at the time of measurement, there is a risk that it will be mechanically damaged by the singulation process, resulting in the deterioration of electrical characteristics and becoming a defective product. Cannot be selected.

  The second method is to measure electrical characteristics after singulation. A semiconductor device that is completely separated as a single piece by removing it from the dicing tape with a picker or the like is used for normal semiconductor measurement. It is a method of carrying and measuring with a handler. According to this method, even if a semiconductor device is mechanically damaged during singulation, resulting in deterioration in electrical characteristics, the semiconductor device can be selected by this inspection. However, as described above, CSP devices are characterized by being small and thin. For this reason, there is a possibility that a conveyance error called a jam frequently occurs due to this conveyance, and there is a risk that the measurement throughput is greatly reduced.

  The third method is to measure the electrical characteristics of a semiconductor device after singulation, but before the semiconductor device is completely separated into individual pieces by the picker, the semiconductor device is stuck on the dicing tape. This measurement is performed (for example, refer nonpatent literature 1). In this method, similarly to the second method described above, even if there is a deterioration in electrical characteristics due to mechanical damage, it is possible to select semiconductor devices having degraded electrical characteristics. Further, although depending on conditions relating to the size of the semiconductor device, it is possible to increase the testing throughput by simultaneously measuring the electrical characteristics of a plurality of semiconductor devices. However, since the positional relationship between semiconductor devices shifts from the design value due to singulation, in order to maintain the measurement accuracy of electrical characteristics, the position of each semiconductor device must be measured accurately, and the electrode position on the measuring device side With respect to the positions of the terminals on the device chip side, the electrical characteristics of the individual semiconductor devices must be able to be measured after making necessary positional adjustments.

  At present, in order to inspect the electrical characteristics of a WLCSP-like semiconductor device or the like, the third method described above is generally used. That is, after the wafer is diced, the inspection is performed in a state where the semiconductor device is stuck on the dicing tape.

In this third method, even if the positional relationship between semiconductor devices is relatively displaced due to singulation, the position of each semiconductor device is accurately detected and the electrical characteristics of the semiconductor device are inspected. Therefore, the present invention relates to a device chip position measurement method.
Joe Sato “Technology Trends of Latest Package Inspection Equipment” Magazine “Electronics Packaging Technology” December 2002 (Vol.18, No.12)

In the method described in Non-Patent Document 1 in which the electrical characteristic inspection of the semiconductor device can be performed by the third method described above, there are problems described later. These challenges
A. Issues related to variations in semiconductor device spacing caused by singulation
A-1. Dicing tape stretch (Simple consideration on maximum stretch)
A-2. Restrictions on multi-site testing Description will be made in the order of the problems related to the alignment of the semiconductor device that has lost teeth during retesting.

A. Issues related to variations in semiconductor device chip spacing caused by singulation Referring to FIGS. 1A and 1B, a process for cutting out WLCSP into individual semiconductor device chips will be described. FIG. 1A is a cross-sectional view schematically showing a state before singulation, and FIG. 1B is a cross-sectional view schematically showing the state after singulation. The WLCSP 10 is affixed on a dicing tape 14 joined to a dicing frame 12. The WLCSP 10 is provided with a portion (dicing line) 16 that is separated into pieces between the device chip 10a and the device chip 10a. In FIG. 1 (A), the portion 16 is displayed in white. A device that divides into individual pieces, such as a dicer, cuts up to a position that reaches the dicing tape or reaches a depth that slightly enters the dicing tape, and separates this portion 16 (hereinafter, sometimes referred to as “single piece”). ).

  When the WLCSP is divided into pieces, the portion 16 to which the device chip 10a is connected disappears, that is, the cutout portion 18 is formed, and the device chip 10a is separated. Further, since the tension of the dicing tape itself in the cut portion 16a is weakened, the dicing tape in the cut portion 16a expands as shown in FIG. Of course, there are other reasons why the dicing tape stretches. For example, the dicing tape expands when a force is applied to extend from the dicing blade. When the dicing tape is extended, the slack 20 is generated due to the weight of the device chip 12a and the dicing tape.

  As described above, when the device chip is singulated, the dicing tape is stretched, which means that the device chip is displaced by singulating from the position where the device chip is regularly arranged before singulation. Means. When a dicer is used to cut a dicing tape until it reaches the position where the dicing tape is present, the separated device chips do not displace independently of each other, but have a loosely connected relationship with the surrounding device chips. Displace collectively while keeping.

  With reference to FIGS. 2 (A) and 2 (B), the state of displacement of the device chip that occurs when the above-described device chip is separated will be described. This figure is a partial schematic drawing seen from a direction perpendicular to the surface of the WLCSP. 2A and 2B, each device chip 10a is indicated by a rectangular thick frame. The street portions 22a and 22b to be cut are linear as shown by 22a in FIG.2 (A) before being singulated, but are straight as shown by 22b in FIG.2 (B) after being singulated. Instead, it tends to be a curve.

  As can be inferred from the arrangement shape of the device chips 10a shown in FIGS. 2A and 2B, the two adjacent device chips are not greatly different in inclination angle with respect to the street between the two adjacent device chips. The interval does not change much compared to before the separation. However, when viewed in detail, the device chip interval is not constant, and the rotation angle of the device chip also varies between individual device chips.

A-1. Dicing tape stretch (Simple consideration on maximum stretch)
As described above, when the WLCSP makes a cut to separate the device chip by the dicer (hereinafter, sometimes referred to as “dicing”), the dicing tape of the cut portion 16a expands. To do. The relationship between the amount of slackness of the dicing tape generated by the expansion of the dicing tape and the size of the maximum amount of extension of the dicing tape will now be considered.

  The inner diameter of a dicing frame for fixing an 8-inch WLCSP is 300 mm. Therefore, with reference to FIG. 3, the case where it is attached to a dicing tape and fixed to a dicing frame having an inner diameter of 300 mm will be taken up as an example. FIG. 3 is a cross-sectional view cut in a direction perpendicular to the surface of the WLCSP affixed to the dicing tape, and abstracts the relationship between the extension amount and the slack amount 28 of the dicing tape 26 mounted on the dicing frame 24. I draw. The amount of looseness is the distance in the direction perpendicular to the non-slack tape surface between the non-slack tape surface of the dicing tape that is not slack and the slack tape surface when it is slack.

For simplicity, it is assumed that the slack portion of the dicing tape has a shape that can be approximated by a part of a circle. In the dicers currently in circulation, the amount of slack allowed for transporting device chips is 5 to 6 mm. Therefore, here, the radius R of the circle is set so that the slack amount 28 is about 10 mm with a margin. Assume a center angle. Assuming that the center angle is 15 °, the angle Θ shown in FIG. 3 is 7.5 °, and the following equation is established.
R × sin (7.5 °) = 150.0 mm
Therefore, R = 1149.0 mm
Extension amount = (R × (7.5 °) −R × sin (7.5 °)) × 2 = 0.858mm
Looseness = R − R × cos (7.5 °) = 9.829 mm
It becomes.

  From the above evaluation, it can be concluded that the amount of slack is about 10 times the amount of tape expansion, and the maximum amount of dicing tape with individual device chips attached is about 1 mm at maximum. .

  FIG. 4 is a schematic view partially showing the arrangement of device chips in a state where the dicing tape is stretched after singulation, as viewed from the direction perpendicular to the surface of the WLCSP attached to the dicing tape. The chip is shown as a quadrilateral surrounded by a thick line. With reference to FIG. 4, the situation in the vicinity of the center portion of the WLCSP and both left and right ends when the maximum extension amount on the dicing tape is assumed to be 1 mm will be described.

  The device chip 30 and the device chip 32 are device chips arranged in different rows adjacent to each other. In FIG. 4, an alternate long and short dash line 34 connects the center lines of the row direction device chips before being singulated. 4 indicates the deviation of the center position of the left end device chip from the center line of the row direction device chip before the singulation of the device chip after singulation.

  If a situation occurs in which device chips in the same row are mistakenly recognized as device chips in other rows, a problem occurs in the device chip test process. However, if the size of the device chip is 2mm x 2mm or more, as shown in Fig. 4, the center line of the device chip near the center in the X direction (left and right direction on the paper) is one at the end of the WLCSP. The fact that it is recognized that it exists at the position of the lower row device chip does not occur from the previous consideration that the deviation max is 1 mm or less.

  As described above, when the size of the device chip is 2 mm × 2 mm or more, the device chip on the WLCSP is overlapped with the row position that is one lower or higher than the row position that designates the device chip. There is nothing wrong with it. Of course, the same can be said about the column direction as well as the row direction.

A-2. Restrictions on multi-site testing
In a test process using a CSP probe (sometimes referred to as a “CSP handler”) that performs electrical characteristic tests of WLCSP, it is common to perform multi-site testing. Multi-site testing means that the test electrodes of the tester are arranged in a matrix by the number to be tested for each device chip, and device tests for a large number of device chips are executed simultaneously. One of the features of the CSP probe is that it can be tested while a large number of device chips are still attached to the dicing tape. By doing in this way, the manufacturing process of a device chip can be shortened.

  FIG. 5A shows a measurement electrode 38 (sometimes called “pogo pin”) of the tester, and FIG. 5B shows a terminal of the device chip 40, that is, device electrode 42 (sometimes called “bump”). .) Is an example. In this example, so-called 2-by-2 multi-site testing is performed in which two testers are arranged in the vertical and horizontal directions orthogonal to each other, and two device chips are arranged in the vertical and horizontal directions orthogonal to each other. Shows the case. Since the number of device chips that can be tested by multi-site testing of 2 rows and 2 columns is 2 x 2 = 4, 2 x 2 is represented as 2 x 2 and multi-site testing of 2 rows and 2 columns. Sting is described as 4 (2 × 2) multi-site testing.

  When device chips are singulated, the device chip interval is generally expanded by being singulated as compared to the interval before the singulation, that is, the design value. With reference to FIG. 6, an example in which the device chip interval is extended and the device chip interval is different from the design value will be described. In FIG. 6, four squares indicate the outer periphery of the device chip 40, and the white circles schematically indicate the measurement electrodes 38 of the tester, and the black squares schematically indicate the device electrodes 42 of the device chip 40.

  As shown in FIG. 6, when the device chip interval is widened by singulation, a shift occurs between the measurement electrode 38 of the tester and the device electrode 42 in multi-site testing. If the test is executed in a state where there is a deviation between the measurement electrode and the device electrode of such a tester, whether the device chip itself as a target of the test is a failure or the contact between the measurement electrode and the device electrode. It is not possible to distinguish whether the failure was determined to be a failure. Further, the device electrode may be damaged by the contact between the measurement electrode and the device electrode that are shifted. That is, if the device chip interval is larger than the allowable value for testing, this device chip must be excluded from being tested.

  With reference to FIGS. 7A and 7B, a coping method when testing a tilted device chip will be described. As shown in FIG. 7A, when the device chip is tilted from the direction in which the measurement electrodes of the tester are arranged, the measurement table on which the device chip attached to the dicing tape is mounted is rotated. As shown in (B), if the angle of the device chip is corrected in the direction in which the measurement electrodes of the tester are arranged, the test can be performed without any problem even when performing multi-site testing.

  Since the testing described with reference to FIGS. 7A and 7B is assumed to be 4 (2 × 2) multi-site testing, the tilt angle of the device chip is the center of two adjacent device chips. The straight line connecting the positions is considered to be the angle formed with the coordinate axis, and is obtained by measurement. When multi-site testing is performed over a range larger than 4 (2 × 2), the center positions of many device chips corresponding to the number of device chips to be measured and the measurement electrodes corresponding to the number of device chips The angle and the parallel movement amount are calculated from the center position of the device unit by using the least square method, and the testing is executed using this value. That is, the method shown in FIGS. 7A and 7B is a method of performing testing after obtaining the position of the device chip and subsequently performing the process of correcting the tilt angle of the device chip. .

  If the inclination angle of each device chip is to be accurately measured, two or more positions must be measured with respect to each device chip. Obtaining the angle of each device chip takes a very long time for alignment, so it is realistic not to incorporate the method of measuring the angle of each device chip into the alignment process during device chip mass production. is there.

  Hereinafter, the term “alignment” is used to determine the position of the device electrode of the device chip in order to align the measurement electrode of the tester and the device electrode when the electrical characteristics of the device chip are evaluated. It is used to mean that the position of the device chip is adjusted. The term alignment is also used to mean that the device electrode position of the device chip is determined and the position of the device chip is adjusted.

B. Issues related to alignment of semiconductor device chips that are missing during retesting Retesting is performed before the picker process or after part of the device chip is removed by the picker after the picker process There are two possible timings. In any case, the tension of the dicing tape changes before and after the measurement due to conveyance by a CSP prober or CSP handler. This means that the device chip interval can change before and after loading the dicing frame into the measuring device (CSP prober or CSP handler).

  Furthermore, in re-testing performed after the picker process, in addition to the change in the device chip interval, the shape of the device chips (two-dimensional array) is the same for each measurement due to the removal of the device chips. Not exclusively. A so-called “tooth loss” condition appears.

  As described above, since the tooth is missing, the device chip existing at the position of the reference measurement point set in the initial testing may be extracted. Under such circumstances, it is necessary to reset the reference measurement points for alignment. In order to obtain the sufficiently high accuracy required in the testing process, it is necessary to consider how the reference measurement point should be reset.

  In the WLCSP testing process, first, each position of the device chip is accurately measured, and the relationship between the position of the device chip and the position of the measurement electrode of the tester is adjusted based on the result, and then the electrical power of the device chip is adjusted. It must be possible to measure mechanical characteristics. However, as described above, in the conventional method, there is a problem to be solved in order to accurately measure the center coordinates of each device chip after being separated into pieces by a dicer.

  SUMMARY OF THE INVENTION Accordingly, the present invention provides a device chip position measuring method that solves the above-described problems and that can measure (calculate) the individual center position coordinates of the separated device chips on a dicing tape. .

  Therefore, the device chip position measuring method of the present invention includes the following steps.

  The first invention is a device chip position measuring method comprising an initial measurement area dividing step and an image processing step.

In the initial measurement area division step, four or more points that do not have three or more points on the same line are set as initial reference measurement points, and a two-dimensional array of device chips is surrounded by the four or more initial reference measurement points. Dividing the initial measurement area. The image processing step measures the center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step, and the device chip center coordinates and the device chip are arranged in a two-dimensional array on the matrix. In this situation, a projective transformation is defined to correspond to an index specified as a combination of two integers indicating the number of rows and columns in which this device chip is located. The center coordinates of all the device chips included in the initial measurement region formed in (1) are calculated using projective transformation .

  In the initial measurement area dividing step, “point” is expressed as a (two-dimensional device chip array) at a position designated as one point on a two-dimensional device chip array (also referred to as an index map). Device chip (in the position specified above as a combination of two integers). The reference measurement point refers to a device chip designated as a point on the index map, and indicates a device chip that measures the coordinates of the center position of the device chip. A measurement region (sometimes referred to as a region) refers to a polygon formed by sequentially connecting reference measurement points set on an index map as region boundary points. All device chips on the WLCSP belong to one of the measurement areas formed at the reference measurement point. The point, the reference measurement point, and the measurement region are common definitions in the following inventions.

  The second invention is a device chip position measuring method comprising an initial measurement region dividing step, an image processing step, a verification step, and a subdivision step.

In the initial measurement area division step, four or more points that do not have three or more points on the same line are set as initial reference measurement points, and a two-dimensional array of device chips is surrounded by the four or more initial reference measurement points. Dividing the initial measurement area.
In the image processing step, four or more device chip center coordinates set as initial reference measurement points are measured, and the position coordinates of the device chip center included in the initial measurement region surrounded by the four or more initial reference measurement points are determined. This is a calculating step.

  The verification step compares the center coordinates of the device chip calculated in the above-described image processing step with the measured center coordinates of the device chip, and the degree of coincidence of both positions is necessary for measuring the electrical characteristics of the device chip. In this step, it is determined whether or not the required matching degree is satisfied.

  The subdivision step is based on the test result in the above-described test step, until the position match level is within the required match level for the measurement region including the device chip that is not within the required match level range. Set the newly selected point as a new reference measurement point so that three or more points do not line up on a straight line, and hierarchically divide the initial measurement area into divided areas surrounded by four or more reference measurement points This is a step to advance the division.

  A third invention is a device chip position measuring method comprising a region data loading step and an image processing step.

  The area data loading step divides the two-dimensional array of device chips into measurement areas surrounded by the four or more reference measurement points, using four or more points that do not have three or more points on the same line as reference measurement points. Thus, information (reference point list data) regarding all the reference measurement points obtained and information (regional data) for managing the relationship between the divided measurement regions and the two-dimensional array of device chips are stored. Loading from a non-volatile storage device.

  The image processing step measures four or more device chip center coordinates loaded as reference measurement points, and uses the regionized data to include devices included in a measurement region surrounded by the four or more reference measurement points This is a step of calculating the position coordinates of the chip center.

  A fourth invention is a device chip position measurement method comprising an initial measurement region dividing step, an image processing step, a reference measurement point coordinate management step, and another reference measurement point selection step.

  In the initial measurement area division step, four or more points that do not have three or more points on the same line are set as initial reference measurement points, and a two-dimensional array of device chips is surrounded by the four or more initial reference measurement points. Dividing the initial measurement area.

The image processing step measures the center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step, and the device chip center coordinates and the device chip are arranged in a two-dimensional array on the matrix. In this situation, a projective transformation is defined to correspond to an index specified as a combination of two integers indicating the number of rows and columns in which this device chip is located. in the heart coordinates of all the device chips included in the initial measurement region formed is a step of calculating using projective transformation.

  The coordinate management step for the reference measurement point is based on this reference position from among the device chips on the dicing tape, with the position arbitrarily set on the dicing frame fixing the dicing tape on which the device chip is placed as a reference position. This is a step of managing reference measurement point coordinates for designating the position of a specific device chip center.

  The different reference measurement point selection step is a polygon formed by connecting the reference measurement points including the other reference measurement point when no device chip exists at the position corresponding to the reference measurement point coordinates, When the device chip cannot be covered, another reference measurement point is selected and added as another reference measurement point.

  According to the first invention, the center coordinates of all device chips singulated on the dicing tape can be calculated. The center position of the device chip in the region surrounded by the initial reference measurement points selected by the initial measurement region dividing step can be obtained by calculation without direct measurement. Therefore, the measurement man-hours can be saved as compared with the method of measuring the total number of device chips and determining their positions.

  According to the second invention, the center coordinates of the device chip calculated in the first invention described above are compared with the measured center coordinates of the device chip. It is determined whether or not the degree of conformity necessary for measuring the electrical characteristics of the device chip is satisfied. Therefore, if the degree of match of the position is within the range of the required degree of match, the process proceeds to the electrical characteristic measurement process. The degree of coincidence is also simply referred to as the degree of coincidence. On the other hand, if the matching degree of the position does not fall within the range of the required matching degree, a subdivision step that is the next step is executed.

  In the subdivision step, based on the test result in the test step, the position match level is required for measuring the electrical characteristics of the device chip for the measurement area that includes the device chip that is not within the required match level range. The initial measurement area is hierarchically divided into the newly selected divided areas surrounded by the four points until the degree of match is within the range. As a result, the degree of coincidence between the calculated center coordinates of the device chip and the measured center coordinates of the device chip falls within the required degree of coincidence, and finally the electrical characteristics of the device chip. You can proceed to the measurement process.

  According to the third invention, information (reference point list data) related to all reference measurement points that are already stored, information (regionalized data) that manages the relationship between the divided measurement regions and the device chip two-dimensional array, Is loaded from the non-volatile storage device, the same effect as the first invention described above can be obtained.

  According to the fourth aspect of the invention, by performing the coordinate management step of the reference measurement point and the separate reference measurement point selection step, it is possible to align the device chip array in the missing state that occurs after the picker process.

  In the coordinate management step of the reference measurement point, set the reference point and reference direction arbitrarily on the dicing frame that holds the dicing tape on which the device chip is placed, and at the time of initial testing and retesting, By adjusting so that the reference point and the reference direction are the same, the center coordinates of the device chip at the reference measurement point are managed. Further, in the separate reference measurement point selection step, when no device chip exists at the position corresponding to the above-described reference measurement point, a new reference measurement point is selected and added as another reference measurement point. Furthermore, even when all the device chips in the region cannot be covered with a polygon formed by connecting the reference measurement points, another reference measurement point is selected and added as another reference measurement point.

  The introduction of the coordinate management step for the reference measurement point and the step for selecting another reference measurement point is to facilitate the alignment in the retesting performed after the picker process. However, for this purpose, the following two items are preconditions. 1). The device chip removed by the picker process is managed by the index map, 2). The change in the device chip interval of the separated device chips mainly occurs during the separation, and the change that occurs when the dicing frame is loaded into the measuring device (CSP prober or CSP handler) It is small compared to the size that is generated during conversion. If the condition 1 is not satisfied, a new reference measurement point cannot be selected in another reference measurement point selection step. If condition 2 is not satisfied, when the dicing frame is loaded into the measuring device (CSP prober or CSP handler), the coordinate value of the reference measurement point measured at the first test is the coordinate measurement step of the reference measurement point. It cannot be used even if it is managed. The fourth invention is for easily performing alignment in retesting performed after the picker process, and is based on the above two conditions.

  In the retesting performed after the picker process, in addition to the change of the device chip interval, the shape of the device chip arrangement (two-dimensional arrangement) is not always the same for each measurement due to the extraction of the device chip, so-called “ Even when the “tooth missing” state appears, the position measurement of the device chip can be performed by performing the coordinate management step of the reference measurement point and the separate reference measurement point selection step. That is, even if a “tooth missing” state appears, it can be obtained with sufficiently good accuracy required in the testing process.

An embodiment of the present invention will be described with reference to the drawings. Each figure shows one configuration example according to the present invention, and only schematically shows the arrangement relationship of each component to the extent that the present invention can be understood. It is not limited to. In the following description, specific devices and conditions may be used. However, these devices and conditions are only one preferred example, and are not limited to these. Moreover, in each figure, the same component is shown with the same number, and the overlapping description may be omitted. Hereinafter, embodiments of the present invention will be described.
1. Alignment of separated device chips
1-1. Region alignment
1-2. Measurement area division method (reference measurement point setting method)
1-3. 1. Alignment procedure of separated WLCSP Alignment of semiconductor device chips in a device chip array in a missing tooth state after the picker process
2-1. Coordinate management of reference measurement points
2-2. Alignment data for retesting
2-3. 2. Alignment for retesting 3. Deviation adjustment method when testing individual device chips 4. Deviation adjustment method and alignment procedure when device chip is large 5. Alignment procedure when device chip is small Description will be made in order along a schematic configuration example of the device chip position measurement system.

1. Alignment of separated device chips The purpose of the alignment process is to align the measurement electrodes of the tester with the device electrodes of the device chip (sometimes referred to as “device chip side terminals”). Is to measure the position of the device chip. At this time, it is important that the error of the mutual positional relationship is equal to or less than an allowable deviation amount (hereinafter, the deviation amount is sometimes referred to as “deviation”) (the measurement electrode coordinates of the tester and the device chip device). It may be called “difference” when compared with the coordinates of the electrodes). That is, it is important that the electrical characteristic measurement for the test is established, and that the mechanical damage does not occur in both the tester side electrode and the device chip side terminal due to the mutual displacement of both electrodes. The allowable amount depends on the shape and size of the device chip side terminal, the shape and size of the tester side electrode, the frequency of the measurement signal, and the number of multi-site testing arrangements.

  In the alignment with respect to the wafer before singulation, the alignment target can be considered as one rigid body, and the alignment is completed when the translational component and the rotational component of the center of gravity of the entire wafer are determined. However, since the device chip interval is not constant in the singulated device chip, it does not mean that the translation component and the rotation component which are the two components are obtained. In each device chip, the translational component and the rotational component change.

  In the testing for the separated strip and WLCSP, as described above with reference to FIG. 6, the measurement electrode and the device chip side terminal were adjusted so as to overlap as much as possible (measured or calculated). ) The distance for parallel movement is controlled according to the center position of the device chip. Further, as described with reference to FIG. 2 or FIG. 4, when there is a slight angle, the control for rotating the measurement table is also required as described with reference to FIG.

  In order to execute such testing, 1) a method of measuring all the device chip center position coordinates is taken, or 2) a plurality of reference device chips are set and the device chip center position coordinates are set. And the center position coordinates of the remaining device chips should be calculated from the center position coordinates of the device chip as a reference (that is, a sampling method). When the method 2) is employed, when the device chip interval hardly changes, the alignment can be completed with the same number of searches as the wafer. The method 1) is not desirable from the viewpoint that the processing time is long. Therefore, the method 2) is specifically devised to show that it is effective as an alignment method for individual device chips.

1-1. Region alignment
Even if the WLCSP is divided into individual pieces, the individual device chips are not displaced independently. As described with reference to FIGS. 2 and 4, adjacent device chips are grouped in relation to each other. Is displaced. As for the position coordinates of a certain device chip center, a sufficiently good approximate value can be calculated from the surrounding device chip center coordinates.

  Here, 16 device chips are arranged in 4 rows and 4 columns, and the center coordinates of all remaining device chips other than the four corners are calculated by measuring the center positions of the device chips existing at the four corner positions. Take up. Hereinafter, the image processing step in the device chip position measuring method of the present invention will be described.

  In the following description, the coordinates are positions in an orthogonal coordinate system with the center of the measurement table as the origin.

<Image processing step>
FIG. 8 is a diagram for explaining the position measurement and interpolation method of the device chips separated in a matrix. In FIG. 8, a thick solid line and a quadrangle indicated by a broken line indicate the outer shape of the device chip, and a thin solid line is a straight line connecting the center positions of these device chips. The device chip indicated by a thick solid line measures the coordinates of the center position of the device chip and designates the position on the array (the array of device chips is a two-dimensional array, and the array of devices is expressed by the array). On the other hand, the device chip indicated by the broken line uses an interpolation method. When a position is designated on the array, the coordinates of the center position of the device chip are calculated.

  In order to calculate the center coordinates of the device chip indicated by the broken line by the interpolation method, the first approximation method is preferably a method using a linear approximation method. In the linear approximation method, when the center coordinates of the device chips located at the four corners arranged in 4 rows and 4 columns are marked as A, B, C, and D, first, the straight line AB, the straight line AC, the straight line BD, and the straight line CD. Find the point that divides. Next, the intersection of the straight lines connecting these points is obtained. The obtained intersection is calculated as the center coordinates of the broken line device chip. Formulas used to calculate the center coordinates of the device chip indicated by the broken line from the positions on the device chip array are expressed by the following formulas (1) and (2).

X = (b ₁ × x + b ₂ × y + b ₃ ) / (b ₇ × x + b ₈ × y + 1) (1)
Y = (b ₄ × x + b ₅ × y + b ₆ ) / (b ₇ × x + b ₈ × y + 1) (2)
X and Y are coordinates of the center of the device chip, and x and y are index positions of the device chip. The coefficients b _{1 to} b ₈ in the equations (1) and (2) are constants that give the relationship between the coordinates X and Y of the device chip center and the indexes x and y that give the index position of the device chip. Since X and Y are the coordinates of the center of the device chip, they take real values. On the other hand, since x and y indicate the index position, that is, in the situation where the device chips are arranged in a matrix, indicate the position of the device chip in each row and column. Take a number.

A set of positive coordinate values x and y indicating the measured coordinate values X and Y of the four points at the four corner positions (center of the device chip indicated by the intersection of the thick solid lines) and the corresponding device chip are expressed by the formula ( Substituting into 1) and equation (2) , the coefficients b _{1 to} b ₈ are obtained. When the coefficients b _{1 to} b ₈ are obtained, if the index position of the device chip other than the four corner positions is designated, the center coordinates of the device chip designated by the index position are obtained.

Next, a description will be given of a position measurement and an interpolation method when device chips are arranged in an incomplete matrix as in the chip arrangement on the periphery of the wafer. One example thereof will be described with reference to FIG. Also in FIG. 9, as in FIG. 8, the thick solid line and the quadrangle indicated by the broken line indicate the outer shape of the device chip, and the thin solid line is a straight line connecting the center positions of these device chips. The device chip indicated by the thick solid line is measured for the center position coordinates and also specifies the index position. On the other hand, in the device chip indicated by the broken line, if the coefficients b _{1 to} b ₈ of the expressions (1) and (2) are known, the coordinates of the center position of the device chip using the expressions (1) and (2) are used. Can be calculated from the index position. In FIG. 9, the rectangle surrounded by the center coordinates A, B, C, and D of the device chip is a figure different from the rectangle. Even in the case shown in FIG. 9, the position of the broken-line device chip can be obtained using the expressions (1) and (2).

Equations (1) and (2) are mathematically known as projective transformation equations. Coefficients b _{1 to} b ₈ correspond to the coordinate measurement values X and Y of the points and the index positions x and y of the device chip by extracting four or more points where three or more points do not line up on a straight line Can be obtained.

  A device chip is designated at the index position for a two-dimensional array representing the arrangement of device chips. When a plurality of device chips are measured, an area on the two-dimensional array of device chips can be surrounded by the designated device chip. In the region defined in this way, many device chips arranged on the wafer can be divided into regions. In other words, by designating the index position of the device chip at the region boundary, the device chip array is divided into regions, and the device chip center coordinates at the region boundary point are measured to calculate the individual device chip center coordinates inside the region. I can do it. Interpolation using equation (1) and equation (2) is an effective way to achieve this goal. In other words, in the approximation method described here, instead of measuring all the device chip center positions, a plurality of reference device chips are designated by their index positions, and the device chip center position coordinates are measured. Then, the center coordinates of the remaining device chips are calculated from the center coordinates of the device chip as a reference (the device chip position to be measured is the boundary point of the region, and the calculated device chip is a point inside the region. It is an interpolation method of the method of making it become.)

  This method will be further described in detail with reference to FIG. Here, the index position of the device chip selected as the reference device chip is referred to as a reference measurement point (here, “initial” is added to be referred to as “initial reference measurement point”). FIG. 10 shows the index position of the device chip that gives the initial reference measurement point on the wafer. The thick solid line, the thin solid line, and the square indicated by the thick broken line drawn inside the circle indicate the device chip. A device chip indicated by a thick solid line provides an initial reference measurement point. An area surrounded by a device chip (initial reference measurement point) indicated by a thick solid line is an initial measurement area.

  When the displacement of each device chip due to the separation into individual pieces is very small and the device chip interval substantially matches the design value (example 1), the device chip indicated by the broken line in FIG. In other words, as a device chip that provides an initial reference point, it is sufficient for alignment to extract the central device chip 0 and the four points 1, 2, 3, and 4 on the upper, lower, left, and right ends. In the testing process, the multi-site testing can be performed by moving the device chip or the measurement electrode based on the device chip interval of the design value. In this case, it is not necessary to divide the measurement area. The alignment method for the wafer can be applied as it is.

  When the displacement of each device chip is small and the device chip interval is slightly different from the design value (example 2), as shown in FIG. 10, the device chip array shown by a thick solid square To divide the whole into areas of a certain size. Then, using the area division point given by the device chip indicated by the thick square as the initial reference measurement point, the individual device chip center coordinates are obtained by calculating from the center coordinates of the device chip at the reference measurement point.

  In FIG. 10, the WLCSP all device chips are divided into nine measurement regions at a total of 16 reference measurement points, 4 each of 1a to 4a, 1b to 4b, 1c to 4c, and 1d to 4d. As can be seen from FIG. 10, the reference measurement point is also a point for dividing the region (region dividing point). In each of these areas, the individual device chip center coordinates are obtained by calculating from the coordinate measurement data of the device chip center position of the reference measurement point. The calculation formulas used at this time are the formulas (1) and (2) already described.

  There are many device chips to be measured that are subject to multi-site testing. In testing, each device chip center coordinate corresponding to the number of chips is calculated from the device chip center coordinates of the initial reference measurement point, and the device chip or measurement is performed as described with reference to FIGS. Control electrode translation distance and angle.

  Using Equation (1) and Equation (2) for situations where device chip spacing changes at a constant rate within a divided region, i.e., increases at a constant rate or decreases at a constant rate. Can respond.

  However, it cannot cope with the increase or decrease of the device chip interval in the region instead of a fixed rate. In such a case, the existing area is subdivided so that the area where the device chip interval increases and the area where the device chip interval decreases are different. Thus, by newly increasing the area and dividing each device chip position into smaller parts within a linear approximation range, it becomes possible to cope with the case where the change in the device chip interval becomes complicated. This step is a subdivision step in the device chip measurement method of the present invention.

  If the projection transformation formulas (1) and (2) are used, as described above, the regionized region does not have to be a rectangle as shown in FIG. In order to use the expressions (1) and (2), it is only necessary to form a region having a region surrounded by four or more points selected so that three or more points do not line up on a straight line.

In the area corresponding to the four corners on the wafer as shown in FIG. 9, it is difficult to make the divided area rectangular, but if using the equations (1) and (2), one such area can be Can be processed as an area. When five or more reference measurement points are used to determine the region, a procedure for obtaining the coefficients b _{1 to} b ₈ using the least square method may be taken.

  Until now, in FIG. 8 and FIG. 9, the boundary point (reference measurement point) of the region has been introduced as previously determined. In FIG. 10, it is described that the region is divided without describing the method of setting the reference measurement point. The measurement area must be divided in the initial measurement area division step before starting the image processing step. The method for calculating the center coordinates of the device chip using the equations (1) and (2), which are the basic part of the present invention, has been described earlier. In the next section, the “measurement area dividing method” will be taken up. .

  To summarize what has been described above, the image processing steps are as follows.

  Pattern matching is performed by moving to the device chip position specified as the reference measurement point for the four reference measurement points in the positional relationship where the three points set in the initial measurement area division step are not aligned on a straight line. Is used to measure the coordinates of the center position of the device chip. When the index position of the reference measurement point is designated at the same time, the coordinates of the center position of each device chip included in the measurement area surrounded by these four points can be calculated.

This image processing step takes the following procedure. That is, the coordinate values X and Y of the device chip center position of the above four points (reference measurement points) are measured, and a set of positive integer values x and y indicating the index position of the corresponding device chip are designated at the same time, Substituting into equations (1) and (2), the coefficients b _{1 to} b ₈ are obtained. When the coefficients b _{1 to} b ₈ of the equations (1) and (2) are obtained, all the device chips existing in the region surrounded by the reference measurement point are obtained by using the equations (1) and (2). The center coordinates X, Y of can be calculated. That is, according to the interpolation method using the equations (1) and (2), the center coordinates X and Y of all device chips other than the device chip that measures the center position of the device chip as the reference measurement points existing in the region are obtained. By designating the index position of the corresponding device chip, it can be calculated in the image processing step.

  The method described above is a very effective method for calculating the center coordinates of all device chips from the measured center coordinates of the device chip at the reference measurement point without measuring the center positions of all device chips on the WLCSP. Yes, significant measurement efficiency improvement can be expected as compared with direct measurement of all device chips.

1-2. Measurement area division method (reference measurement point setting method)
The device chips arranged on the wafer can be represented in a two-dimensional array. It is convenient if the position of each device chip on the dicing tape with the WLCSP attached can be specified easily. For this purpose, an index map is used. Device chips are two-dimensionally arranged on a dicing tape to which WLCSP is attached. A diagram in which individual device chips are associated in a two-dimensional array (two-dimensional index) is an index map. FIG. 11 shows an example of the WLCSP index map. A coordinate axis is provided in each of the X-axis direction in which the right direction is a positive direction and the Y-axis direction in which the upward direction is a positive direction, and an integer is associated with each device chip. In FIG. 11, squares are drawn at positions corresponding to all sets of integers in the X-axis and Y-axis directions. Among these, the square indicated by the solid line indicates the position where the device chip actually exists, and the square indicated by the broken line indicates the position where the device chip does not actually exist. When the position of the quadrilateral shown by the broken line in the lower left is designated as the reference position by (1, 1), each device chip in FIG. 11 can be designated by a combination of two integer values. The combination of these two integer values is called a coordinator coordinate, and such a coordinate system is called a coordinator coordinate system.

  Since the wafer is circular and the device chip is rectangular, there are no device chips at the upper left, lower left, upper right, and lower right corners of the WLCSP index map. The area where the devices around the four corners at the upper left, lower left, upper right, and lower right do not exist varies depending on the size of the device in the horizontal and vertical directions (device size). The WLCSP index map is divided into regions by the length (number of devices) in the horizontal and vertical directions. When the image is divided into regions, as shown in the image processing step, when the center position of the device chip existing at the region dividing point is image-measured, the remaining device chip center coordinates in the region are obtained by calculation. A point to be divided into regions (region dividing points) is referred to as a reference measurement point in this sense. Here again, it is described that the reference measurement point (area division point) is a point on the index map (a point indicating a device array).

  Referring to FIG. 12, consider the overall trend of device chip displacement. The WLCSP before being singulated is shown in FIG. 10 (in FIG. 10, one device chip is represented by a square, and the device chip is not displaced at all). When separated into pieces, the position of the device chip is displaced. The situation is shown in FIG. 12 (although device arrangement is not the same in FIG. 10 and FIG. 12). FIG. 12 shows the arrangement of device chips (how the device chips are displaced). Precisely, the size of each device chip does not change, and the situation shown in FIG. Dividing the entire device chip array shown in Fig. 12 into appropriate areas and measuring the center position of the device chip that is the boundary point of each area, all the device chip center coordinates within the area are calculated. Has already been described in the image processing step.

  As an example of area division in FIG. 12, the device chip at the position marked with + is divided as an area division point (initial reference measurement point). In FIG. 12, a cross mark (+ mark) indicates the center position of the device chip that determines the initial reference measurement point. A plurality of regions surrounded by broken lines indicate regions formed by division, that is, measurement regions. For example, FIG. 12 shows an example of the measurement area obtained by dividing the area B indicated by diagonal lines. In FIG. 12, the WLCSP is deformed so as to pull the four corners at the upper left, lower left, upper right, and lower right as a whole (deformed into a constricted pincushion at the center). The WLCSP thus deformed was divided into regions at the position of the + mark. In each region of FIG. 12, it can be said that the direction and magnitude of the displacement of the device chip can be adjusted by a first-order approximation.

  The displacement of the device chip on the WLCSP depends on the singulation process. If the singulation process is the same, the overall tendency of the displacement of the device chip on the WLCSP can be estimated to be almost the same. This is because when dicing, the dicer blade acts to shift the device chip, the device chip counteracts the shifting action due to the adhesive force with the tape, and the adhesive force depends on the contact area between the device chip and the tape. It is proportional. Therefore, the magnitude of displacement of the device chip depends on the size of the device chip and also on the cutting order.

  FIG. 12 shows an example of the overall tendency of the device chip displacement. If the conditions that cause the displacement of the device chip are the same, it is estimated that the overall tendency of the displacement is almost the same. That is, the conditions that cause the displacement of the device chip are the device type and the individualization process.

  Since the same device type is the same individualization process, a precondition is set that “the overall tendency of the displacement of the device chip is the same when the device type is determined”. In this method, the center coordinates of the device chip are obtained by dividing the WLCSP into regions, and the region dividing points of the WLCSP reflect the overall tendency of the device chip displacement. If such conditions are provided, it is possible to refer to and use WLCSP area division points (standard measurement points) in the same type of testing thereafter. The condition that the overall trend of device chip displacement will be the same when the device type is determined is set. However, when a large number of WLCSPs are tested, the set WLCSP area division points may not be sufficient. Yes (the tendency of the displacement of the device chip varies). To cope with such a situation, there are a verification step and a subdivision step. When the measurement area is subdivided based on the result of the verification step, it is divided into more suitable areas.

  The region division points in FIG. 12 are appropriately arranged and divided into regions in the figure, but the procedure for determining the reference measurement points for logically implementing the region will be described as the initial measurement region division step. How much the center position of each device chip calculated from the center coordinates of the reference measurement point is shifted cannot be determined unless the measured center coordinates are compared with the calculated center coordinates and checked. The verification step is a procedure for checking the calculated center coordinates of each device chip and the measured center coordinates of the device chip and confirming that the difference is not more than an allowable value. In the comparison check of the verification step, it was determined that the difference between the measured values and the values calculated using the projective transformation formulas (1) and (2) was greater than the allowable value for the coordinates of the individual device chip center positions. In some cases, it is necessary to divide the measurement region further smaller (subdivide the region).

  The subdivision of the area is performed only for the area where the above-described difference is determined to be greater than or equal to the allowable value. The procedure for subdivision of the region is a subdivision step. In the subdivision step, new reference measurement points are selected and added. In the following description, the initial reference measurement point is limited to the initial value, but after dividing the area again and selecting a new reference measurement point, the initial measurement point is referred to as the reference measurement point. It may be read as: Therefore, in the following explanation, it expresses without attaching the limitation of the initial stage.

  In the measurement area division method (reference measurement point setting method), the initial area division (initial reference measurement point and initial measurement area) is first obtained in the initial measurement area division step, and the center position of each device chip is calculated in the verification step. The obtained result is compared with the actual measurement value to check whether the initial region division is good. A re-dividing step for re-dividing the area is performed on the area determined to be NO.

<Initial measurement area division step>
As a procedure 1, an initial value of the number of area indexes that gives a minimum interval between device chips used for area formation is set. A region index number of 1 means that the space for one device chip is the space between regions. An area index number of 5 means that an interval corresponding to five device chips is used as an area interval. In the procedure 1, the initial value of the index number of the device chip used for the region is determined for each of the X and Y directions from the device chip size and the overall size used for the region.

  The process of dividing into regions will be described with reference to FIG. In FIG. 13, the positive direction (row direction) of the X axis is rightward, and the positive direction (column direction) of the Y axis is downward. In FIG. 13, the number of regional indexes is 5 in both the X and Y directions. That is, the interval between five device chips in both the X and Y directions is set as the region interval. If this index is 1, it means that one area is composed of one device chip in this case, which means that all device chips are to be measured.

  The maximum number of regional indexes is an integer part obtained by dividing the maximum value of the number of device chips arranged in a row by 2. That is, the maximum value of the number of device chips arranged in a line is the maximum number of device chip rows and / or column arrangements. For example, since the maximum number of device chips on the horizontal line indicated by the arrow in FIG. 13 is the maximum number, in the example shown in FIG. 13, the maximum value of the number of device chips is 31. Therefore, the maximum number of regional indexes is 15.

  As step 2, determine the central device chip. The index position of the central device chip is determined from the device chip array. If the index array is an even number, which of the two device chips located at the center is to be determined as the central device chip is determined in advance, and the central device chip is determined accordingly.

  In step 3, a device chip existing at a position determined by the central device chip and the number of regionalized indexes or an integer multiple of the number of regionalized indexes is designated. Based on the central device chip, a device chip that includes the central device chip itself and is positioned for each of the number of area indexes in the X and Y directions given in step 1 is selected as the reference device chip. When the (number of device chips) / (regional index) number is an odd number, the central device chip is set to the center of the region. In FIG. 13, this reference device chip is indicated by a thick quadrilateral. One of these thick quadrilaterals is represented by the symbol c. Similar to the device chip to which the symbol c is attached, all device chips indicated by thick squares are device chips selected as reference device chips. What is defined as a reference device chip here is synonymous with what is described as a reference measurement point.

  In step 4, a device chip existing on the outer edge of the device chip array is designated as a reference device chip. There is a case where the reference device chip does not exist at a position determined by the number of regionalized indexes or an integer multiple of the number of regionalized indexes with respect to the central device chip (index position outside WLCSP). In this case, the outer edge device chip of the device chip array is designated as the reference device chip. In FIG. 13, such a reference device chip is represented by a quadrilateral surrounded by the second thickest line segment. A symbol d is representatively shown for one of the quadrilaterals surrounded by the second thickest line segment. Similar to the device chip to which the symbol d is attached, all the device chips indicated by the quadrilateral surrounded by the second thickest line segment are the outer edge device chips of the device chip array designated as the reference device chip.

As step 5, complete the basic area work. The area determined up to step 4 is called a basic area. In the example shown in FIG. 13, there are six regions in both the vertical and horizontal directions. These area numbers are represented by (x _B , y _B ). Here, x _B and y _B are positive integers from 1 to 6. For example, in FIG. 13, the reference area indicated by hatching is at a position twice as many as the number of regionalized indexes in the X-axis direction and three times as many as the number of regionalized indexes in the Y-axis direction. The area number is (2, 3).

As step 6, the following items are inspected for the reference device chip selected so far (synonymous with the region boundary point and the reference measurement point). For each area number (x _B , y _B ) of the basic area, it is determined whether each device chip position other than the reference device chip is inside or outside the polygon formed by connecting the area boundary points. Specifically, the region boundary points are rearranged clockwise or counterclockwise, and then the region boundary points are rotated once for each device chip position in the region (a position other than the region boundary point). Determine if it is inside the area.

  A method for determining the inside and the outside of the area will be described with reference to FIG. In FIG. 14, as an example for some of the regions in FIG. 13, from left to right, region number (1, 2), region number (1, 3), region number (2, 1), region number (2, 2 ), The basic area given by the area number (2, 3). A trajectory that makes one round of the region boundary point is indicated by a thin broken line, and a device chip e existing at a position that is not determined to be inside the region is indicated by hatching. A method for determining whether the point P is inside or outside the quadrilateral will be described with reference to FIGS. 15 (A) and 15 (B).

15A, when the point P at the position indicated by the cross exists inside the quadrilateral 1234, FIG. 15B shows the point P at the position indicated by the cross exists outside the quadrilateral 1234. Each case is shown. The angle of the side 12 with respect to the point P is A ₁ , the angle of the side 24 with respect to the point P is A ₂ , the angle of the side 43 with respect to the point P is A ₃ , and the angle of the side 31 with respect to the point P is A _4. Add together with the sign. As shown in FIG. 15A, when the point P is inside, the above sum A ₁ + A ₂ + A ₃ + A ₄ is 2π, and when it is outside, the sum is 0. Become. As a special case, when the point P is on the boundary line, this sum is π. In this way, the position of the device chip e indicated by hatching in FIG. 14 is determined to be outside the region.

  If all the device chip positions are inside the area, step 7 is executed. On the other hand, if there is a device chip outside the area, step 8 is executed. Whether the case on the boundary line is internal or external depends on the overall judgment, but is internal in the flow of FIG.

  As step 7, the shape of the region formed by the region boundary points (polygon formed by the region boundary points) is examined. When the polygon formed by the region boundary points is a quadrilateral or more, the region boundary points of this region always include four points that are not on the same straight line. Further, when there are five or more area boundary points, projective transformation parameters are obtained by the least square method. In the special case where the outer shape formed by the region boundary points is a triangle, a line segment, or a point, the correspondence is changed as follows. When the outer shape of the area is a triangle like the area given by the area number (1, 6) shown in FIG. 13, the position of the device chip existing inside the area is the device chip existing at the apex position of the triangle. It is determined by performing weighted average calculation using coordinates.

  Also, when the outline formed by the area boundary points is a line segment, the position of the device chip existing inside the area is calculated by performing a weighted average calculation using the device chip coordinates existing at both end positions of the line segment. We will ask for it. If the region is a point, the device chip at that point is measured. For cases where the outline formed by the region boundary points is a triangle, line segment, or point, the special treatment is that the projective transformation using equations (1) and (2) is defined by 8 parameters. This is because, as a condition for obtaining these parameters, a condition that three of the four points are not on the same straight line is necessary.

  As step 8, region boundary points are selected from device chips existing outside the polygon formed by the region boundary points and added. When the device chip is outside the polygon formed by the region boundary points, the region boundary points are added. When there are a plurality of points on the outside, the region boundary point (reference measurement point) is first added with one point, and whether or not all device chip positions belonging to this region are present internally, Check again with the method described with reference to FIG.

  The selection of one point to be added is performed by the following method. A method for adding a new region boundary point will be described with reference to FIG. In FIG. 16, each device chip is indicated by a square. First, the region boundary points are arranged clockwise. The distance between the point of the device chip determined to be outside and the line segment connecting the two region boundary points rearranged is obtained. The line segments are changed in order, and the minimum distance is set as the discrimination distance of the device chip. A line segment connecting the end region boundary point to the first region boundary point is also included in this. For each device chip determined to be outside, this discrimination distance is obtained, and the device chip having the maximum discrimination distance is selected as a region boundary point to be newly added. FIG. 16 shows a device chip f which is filled with fine horizontal line segments to give this newly selected region boundary point to be added. A line segment indicated by a solid line in FIG. 16 is a line segment for obtaining the discrimination distance.

The procedure described above is shown as a flowchart in FIG. In step S1, a procedure 1 for setting an initial value of the number of indexes for area conversion is executed. In step S2, procedure 2 for determining the central device chip is executed. In step S3, a procedure 3 for designating the central device chip and the device chip at the position of the number of regionized indexes is executed. In step S4, if there is no device chip at the position to be specified as a result of executing step S3, step 4 is executed to specify a device chip that supplements the region index number positions in the X and Y directions. In step S5, the step 5 of completing the basic region forming operation and assigning the region number (x _B , y _B ) to the region formed by the basic region forming operation is executed. In step S6, step 6 is executed to determine whether or not all device chip positions other than the region boundary points are inside the polygon formed by the region boundary points.

  This determination is executed in step S7, and if all device chip positions other than the region boundary point are inside, the process proceeds to step S8, and if there is an external one, the process proceeds to step S9. In step S8, the above-described procedure 7 is executed. When the region outline formed by the region boundary points is a triangle, a line segment, or a point, special measures are taken when obtaining the center coordinates of the device chip existing inside the region. For example, when the outer shape of the region is a triangle, the center coordinates of the device chip existing inside the region are obtained by performing a weighted average calculation using the device chip center coordinates existing at the vertex position of the triangle. In this case, the interpolation calculations (1) and (2) are not used. In step S9, the above-described procedure 8 is executed. That is, the region boundary point is selected from the device chips existing outside the polygon formed by the region boundary point and added.

  Steps S6 and subsequent steps are executed until the steps up to the procedure 8 described above are completed for all the basic regions, and the process shown in this flowchart is terminated when the steps are completed.

  By performing steps 1 to 8 of the above-mentioned regionization, the regionization and region boundary points can be determined as shown in FIG. That is, a region boundary point (reference measurement point) for determining the divided region can be determined.

  FIG. 18 is a diagram showing a state immediately after the area formation described with reference to FIG. 13 is completed. The peripheral part device chip of the wafer indicated by a broken line in FIG. 18 (typically, a corresponding device chip is given a symbol g) is a region boundary point (reference measurement) added by executing step 8. A device chip giving a dot) is shown. When the regionalization is completed, the regionalized data is stored in a nonvolatile storage device to be described later. Since the regionalized data at this stage may be rewritten (the test has not been completed), it is stored as “save candidate”. The reference measurement point is also a region boundary point. Therefore, one reference measurement point may belong to a plurality of areas. It is convenient if the regionized data is organized by region. On the other hand, with respect to the reference measurement point, one reference measurement point may become data of a plurality of areas, but it is convenient to register so as not to overlap. Therefore, the area data and the reference measurement point list data are stored as part of the parameter set of the product parameters.

As regional data,
Regiond data header test data (ie, allowable deviation) X, Y
Total number of areas N
Data part of regionized data 0 to N-1
Regionalized data number n (serial number)
Attribute id indicating the validity of invalidation data
Area number (X direction, Y direction) x _B , y _B
Start index number (X direction, Y direction) ix, iy
Number of indexed regions (X direction, Y direction) jx, jy
Number of reference measurement points in the area M
Reference measurement point number in the region n ₀ ... n _M-1 (Specify with reference measurement point number: M)
Manage in the form of As reference measurement point list data,
List of reference measurement points Data header part All reference measurement points K
Data part of reference measurement point list data 0 to K-1
Reference measurement point number k (serial number)
Attribute indicating reference measurement point valid / invalid kd
Index position (X direction, Y direction) kx, ky
Neighboring device chip presence / absence data kc
Number of reference areas J
Reference area data number k ₀ ... k _J-1 (Specified by area data number: J)
Manage in the form of The adjacent device chip presence / absence data is used in the process of taking correspondence between the index map and each device chip (in the alignment procedure) and in the case of a small device chip. Details will be described later.

  In general, when the device chip size is large, the displacement of the device chip due to singulation is small, but when the device chip size is small, the displacement of the device chip due to singulation is large. Therefore, for WLCSP with a large device chip size, there is no problem in many cases even if the alignment of the device chip is started based on the area data obtained by the above procedure. However, how much the individual device chip center positions calculated from the device chip center coordinates of the reference measurement point are not determinable unless actually measured. The verification step is to check the calculated center coordinates of the individual device chips and the center coordinates of the actually measured device chips to confirm that the difference is not more than an allowable value.

<Examination step>
It moves to the index position picked up as a reference measurement point, and the coordinates of the device chip center position are measured by pattern matching using image processing. Then, the device chip center coordinates of all the reference measurement points are sequentially measured. When the device chip center coordinates of all reference measurement points are measured, all device chip center coordinates in the region are calculated in units of measurement regions (finally, all device chip center coordinates are calculated). The process so far is the image processing step already described. In step 9, the device is moved to all device chip center positions on the WLCSP (the calculated device chip center coordinates are moved as guide positions), and the device chip center coordinates are measured. In step 10, the calculated center coordinates of each device chip and the measured center coordinates of the device chip are checked, and it is confirmed whether or not the difference is equal to or less than an allowable value. Procedure 10 is a verification step in the device chip position measurement method of the present invention.

  In the verification step, the difference between the measured coordinate value and the coordinate value calculated using the projective transformation equations (1) and (2) is greater than or equal to the allowable value for the coordinates of the center position of each device chip. If it is determined, the measurement region including the device chip for which the difference is determined to be greater than or equal to the allowable value is further divided (redividing the region). When the measurement region is subdivided, it is necessary to perform the verification step again for each subdivided region. However, in this state, the center coordinates of all device chips have already been measured in step 9. Therefore, the procedure 9 (measurement of center coordinates of all device chips) is omitted in the verification step for the subdivided area. Subdivide the area in step 11. The subdivision step will be described later.

  When the area is subdivided, new reference measurement points are added. The device chip center coordinates of the new reference measurement point can be copied because the center coordinates of all device chips have already been measured in step 9 (step 12). From the device chip center coordinates of the reference measurement point of the subdivided area, the center coordinates of all device chips belonging to the area are calculated (procedure 13). When the image is subdivided, there will be a plurality of regions before the subdivision. In order to verify the center coordinates of all the device chips belonging to the region before being subdivided, the number of times the region is divided is repeated (procedure 14). By repeating the image processing step and the above-described procedure 9 to procedure 14, the regionized data and the reference measurement point list data to be a definite version are obtained. The meaning of the final version indicates that the measured value and the calculated value are confirmed to be within the allowable value.

  FIG. 19 shows a flowchart for explaining the procedure of the verification process. With reference to the flowchart shown in FIG. 19, step 9, step 10, step 11 (region subdivision process), and the subdivision region verification process will be described. In step S90, steps 1 to 8 of the initial measurement area dividing step described above are applied and executed to calculate a reference measurement point. This step S90 is an initial measurement region dividing step in the process (S1 to S10) already described with reference to the flowchart shown in FIG.

  In step S91, the device chip is moved to the reference measurement point device chip, and the device chip center coordinates are measured by pattern matching using image processing. When the device chip center coordinates of all the reference measurement points are measured, all device chip center coordinates belonging to the area are calculated from the device chip center coordinates of the reference measurement point for each initial measurement area (step S92). Two steps, step 91 and step 92, are image processing steps. In step S93, the device moves to all device chip center positions on the WLCSP, and the device chip center coordinates are measured by pattern matching using image processing (the measurement method is the same as the reference measurement point).

  It may take a long time to measure all device chip center coordinates. In such a case, sampling is performed. In step S94, the measured values of the center coordinates of the device chips and the values calculated from the equations (1) and (2) that give the projective transformation are checked for all device chips. Since the measurement area may be subdivided, the initial measurement area is used as a unit, and all device chips included in the area are examined. It is determined whether or not the difference (deviation) between the measured value and the calculated value with respect to the center position of the device chip in step S94 is within an allowable value (step S95).

  As a result of this determination, if it is within the allowable value, it is confirmed that this work is finished in all the initial measurement areas, and all the work is finished (S99). On the other hand, if it is determined that this difference (deviation) is outside the allowable range, the region is further subdivided (S100). The device chip center coordinates of the reference measurement point newly set at the time of subdivision are copied (step S98), and the center coordinates of all the device chips to which the area belongs are obtained from the device chip center coordinates of the reference measurement point for each area. Calculate (step S97). Since the process of dividing the area is included, when the area is divided in units of the initial measurement area, the divided number is repeated (step S96). The processes of S100, S98, S97, and S96 are repeated until the difference (deviation) between the measured value and the calculated value is within the allowable range.

  The reference measurement point that has completed the process shown in the flowchart of FIG. The area data and the reference measurement point list data are stored in a nonvolatile storage device to be described later.

  In a practical program for executing the above-described procedure, it is ideal that it is guaranteed that one area can be divided and processed so that one area is constituted by one device chip. However, even when performing an actual work with such an ideal program, it is preferable to define the minimum number of indexes to be defined as about 2 or 3 and execute the above-described procedure.

  The fact that it is necessary to proceed with the division in order to make the region into a fine level means that the device chip position variation as shown in FIG. If every other or two device chips are measured and the device chip position between them cannot be approximated, multi-testing should be considered fundamentally difficult. In actual operation, it is possible to determine to what extent multi-testing is performed by giving the minimum number of indexes to be regionized.

  By the way, the calibration process is based on measuring the positions of all device chips. When the device chip size is small, there are a considerable number of device chips on one WLCSP wafer. When the device chip size is small, in order to shorten the time required for the process in the verification process, it is important to measure not to measure the positions of all device chips. When every other or two device chips are measured, the number of measurements is greatly reduced.

  At the position close to the outer periphery of the WLCSP, there is a high possibility that the displacement of the device chip due to singulation is high, so that the option to add to the measurement can be selected for all positions of the device chip close to the outer periphery (specify the number from the outer periphery) To do. Since such sampling is simple, details are omitted.

  Since the center coordinates of all device chips are measured by image processing, it takes a long time to execute the verification process, but the longest time is about the same as the time required to measure all device chip positions. The test need not be performed every time. When testing a new device chip type, it is desirable to perform several tests. If the test fails continuously (determined as a defective product), there may be a gap between the device chip terminal position and the tester measurement electrode position. In some cases, it is desirable to take care so that the test can be performed.

<Subdivision step>
If the difference between the measured value of the device chip center coordinates and the device chip center coordinates calculated using the projective transformation equations (1) and (2) is greater than or equal to the allowable value, the measurement area is subdivided. Only the region where the above-described difference is determined to be greater than or equal to the allowable value is performed. The subdivision of the area is performed according to the flowchart of FIG. The flowchart of FIG. 20 is similar to the flowchart shown in FIG. The regionized data and reference measurement point list data stored as “candidate storage” are called (step S80). A new reference measurement point is added to an intermediate point of the reference measurement points set in the area to be subdivided, with the number of area indexes in the area to be subdivided being halved (step S81). When the region to be re-divided includes the outer edge device chip of the device chip array, the outer edge device chip of the device chip array located in the same row or column as the reference measurement point added in step S81 is added as the reference measurement point. (Step S82). Thereafter, the area to be subdivided is divided into four areas (step S83). The process after the division into four areas is the same as the procedure after the procedure 6 of the initial measurement area division step.

  With reference to FIG. 21, a situation where a newly added reference measurement point and one area are divided into four will be described. In FIG. 21, a device chip is indicated by a square, and a device chip indicated by a thick square among the squares is a device chip that provides a reference measurement point (region boundary point).

  Here, the example described with reference to FIG. 21 is a case where the central region (regions indicated by reference measurement points P, Q, R, and S) is further divided. The management of the area is performed using the area data. For the newly subdivided area, new area data is added. An attribute for invalidating the regionized data of the region to be divided is attached. The region numbers constituting the regionized data are based on the region numbers determined by the basic regionization (procedure 5 in FIG. 17), and are managed by adding branch numbers when the regions are divided. If managed in this way, the corresponding regionized data can be easily referred to for the device chip position for which the coordinates are to be obtained. As shown in FIG. 21, the reference measurement points (region boundary points) added on the region boundary line belong to a plurality of regions. In FIG. 21, the added reference measurement points are indicated by T, U, V, W and X.

  The added reference measurement points T, U, V, and W belong to each of the four areas generated by the subdivision, but each of the upper, lower, left, and right areas of the subdivision area also has. belong to. Since the area data includes the start index number and the area index number, the position and range of the area can be easily understood. It is easily determined whether the added reference measurement point is a region boundary point of the region to be inspected. If it is included in the area, the reference area data number of the standard measurement point list data is also corrected. Consistency between the regionized data and the reference measurement point list data is ensured.

1-3. Alignment Procedure for Individualized WLCSP The alignment procedure for the segmented WLCSP will be described. When WLCSP is singulated, the intervals between individual device chips are not constant. Furthermore, each device chip has a slight angle as shown in FIG. 2 or FIG. However, in order to perform multi-site testing, even a singulated device chip cannot be multi-site tested by a CSP probe (CSP handler) unless it is aligned to some extent.

  The points to be considered when performing alignment are summarized as shown in the following 1) to 3).

  1) The device chip interval is not constant. Multi-site testing cannot be performed if there is some variation in device chip spacing. Therefore, when the device chip interval changes by more than a specified allowable value from the design value, an error condition is taken into alignment.

  2) If the device chip interval changes by more than the specified tolerance from the design value due to the separation, if an error condition is introduced, all device chip intervals on the WLCSP must be measured. This must be done, but this is not realistic. Therefore, as shown in FIG. 10, the WLCSP is divided into regions, and the region dividing points (reference measurement points) are measured. The device chip interval is checked in the alignment process.

  3) Although the device chip is tilted to some extent as shown in FIGS. 2 and 4, it is not practical to measure the tilt angle of each device chip (in order to measure the tilt angle of each device chip, Two or more locations must be measured for one device chip, and all device chips must be measured). Therefore, the inclination angle of the device chip is calculated from the center positions of a large number of device chips corresponding to the number of multi-site testing. As shown in FIGS. 6 and 7, the actual device chip tilt angle correction is executed by controlling the measurement table in the testing process.

  Furthermore, as already discussed, the maximum extension of the dicing tape may be considered to be 1 mm or less. When the maximum extension amount of the dicing tape is 1 mm or less, if the size of the device chip is 2 mm × 2 mm or more, the center portion and the both end portions of the WLCSP as shown in FIG.

  In the separated WLCSP, the angles of the individual device chips are not the same, and the intervals between the device chips are not the same. After the alignment of the WLCSP is completed, the subsequent testing corresponds to the measured position of each device chip, and the correction operation as described with reference to FIGS. 6 and 7 (device chip measurement table). Move it up). Considering the operation for correction, the alignment process measures the average device chip angle, adjusts this angle, and then measures the center coordinates of the device chip with reference to this angle, and To do. As described above, the center positions of all device chips are obtained by measuring the center coordinates of the device chip at the reference measurement point (area division point).

  A process for performing alignment of the separated WLCSP will be described with reference to a flowchart shown in FIG. Load the regionalization data and reference measurement point list data stored as part of the parameter set of product parameters. In the case of testing for the first time with the same product type, the initial measurement area dividing step is executed to create aread data and reference measurement point list data. The loaded data is data used for only one WLCSP to be tested as work data. This process is step S30 in the flowchart shown in FIG.

  When the WLCSP attached to the dicing tape is loaded onto the measurement table, the WLCSP is inclined at an angle as a whole. The reason why the whole WLCSP is tilted at one angle is that the angle deviation occurs when the WLCSP is affixed to the dicing tape, and the angle deviation occurs when the dicing frame is loaded on the measurement table. There are two things to do. In the flowchart shown in FIG. 22, step S31 displayed as “Adjusting the angle of WLCSP placed on the measurement table” corresponds to a step of measuring and correcting this angle.

  The process shown in the flowchart shown in FIG. 22 will be described with reference to FIG. FIG. 23 is a diagram for explaining the process for adjusting the angle of the WLCSP. The outer periphery of the wafer is shown as a circle including the orientation flat, and the device chip is shown as a fine square.

  In the process of correcting the angle, the first point is the pattern matching of the center position of the device chip (device chip shown as 1 in FIG. 23) existing near the center of the WLCSP or the position away from the center of the device chip. Measurement is performed by a method applying image processing such as (synonymous with measuring the center coordinates of the device chip). Subsequently, a second point (a device chip indicated as 2 in FIG. 23) that is several device chips away from the first point is measured by the same method, and an angle is calculated from the direction in which the first point and the second point are connected. As a special case, the calculated angle may be corrected at this stage. You may correct | amend immediately after calculating | requiring an angle each time after that.

  The third method is measured using the same method in the vicinity of the right edge of the WLCSP (device chip indicated by 3 in FIG. 23) on the extended line in the direction connecting the first and second points. Determine the straight line connecting the 1st, 2nd, and 3rd points using the method of least squares. A fourth point is measured using the same method in the vicinity of the left end of WLCSP (device chip indicated as 4 in FIG. 23) on the extension of this straight line. Determine the straight line connecting the 1st, 2nd, 3rd and 4th points using the least squares method. An angle is obtained from the slope of the line thus determined, and this angle is corrected as the slope of WLCSP.

  Even if the angle of the WLCSP is corrected as described above, the WLCSP is still shifted to the state of being translated with respect to the measurement table due to the following two factors. This deviation occurs when the WLCSP is affixed to the dicing tape or when the dicing frame is loaded onto the measurement table. The magnitude of the deviation due to the above factors also depends on the device used in the previous process. For this reason, a process for taking correspondence between the index map and each device chip (step S32) is required.

  Referring to FIG. 24, the process for taking this correspondence, that is, step S32 will be described. FIG. 24 is a diagram for explaining the process of taking the correspondence between the index map and each device chip. The outer periphery of the wafer is indicated by a circle including an orientation flat, and the device chip is indicated by a fine square.

  First, a specific position on the device chip arrangement is set, and a device chip corresponding to this is searched to confirm its existence. In FIG. 24, the device chip at this specific position (indicated by a thick square) is indicated by A to D. The device chip at the specific position is selected from the reference measurement points. The selection method is to select four lines inclined at an angle of 45 degrees as the first contact point when approaching from the outside of the WLCSP (if there are multiple lines, select the top). The arrangement of device chips existing around the points A to D in FIG. 24 is also important. The arrangement of the device chips existing around is encoded by the method shown in FIG. 25 and becomes the adjacent device chip presence / absence data in the reference measurement point list data.

  FIG. 25 is a diagram illustrating a method of coding the presence / absence of adjacent device chips. When the device chip is present, 1 is indicated, and when the device chip is not present, 0 is expressed as a binary number in order from the upper left position, and the result is a numerical value. In A of FIG. 24, 00111000 becomes 56. The “adjacent device chip presence / absence data” of reference measurement points other than A, B, C, and D is set to −1. This setting is performed at the stage when the initial measurement area dividing step is completed and the reference measurement point candidates are obtained.

  In the set A to D, the presence / absence of a device chip is checked to determine whether the surrounding device chip matches the adjacent device chip presence / absence data. The presence / absence of the device chip is confirmed by the same pattern matching method as that for measuring the center coordinates of the device chip. Assume that the correspondence between the index map and each device chip is obtained when three locations A to D match. That is, step S32 is ended.

  When correspondence with the index map is obtained, the device chip center coordinates of all the reference measurement points in the reference measurement point list data are measured by image processing by pattern matching. Measurement is performed by moving one point at a time to the center position of the device chip at the reference measurement point (step S33). In step S34, it is checked whether the center coordinates of the device chip at the reference measurement point have been measured. When the device chip center coordinates of all the reference measurement points are measured, the center coordinates of all the device chips included in the region are calculated by using the area data, using Equation (1) and Equation (2) (Step S35). ). If the center coordinate of the device chip at a certain reference measurement point is not measured, an alternative reference measurement point for that reference measurement point is set and added to the reference measurement point list data (step S36). Then, the reference measurement point in the next order is measured. Steps S33, S34, and S35 are image processing steps.

When there is no verification step, this is the end of alignment. That is, since the center coordinates of all the device chips are obtained in step S35, the testing process is started using the coordinates. The center coordinates of the obtained device chip are the same two-dimensional array data as the index map. The center coordinate data of the device chip is named the measurement alignment data, the index position is (i, j),
Measurement alignment data header part Number of arrays (X direction, Y direction) Kx, Ky
Data part of measurement alignment data Attribute id [i] [j] indicating device chip valid / invalid
Center coordinates of device chip (px [i] [j], py [i] [j])
Are stored in a nonvolatile storage device if necessary. The device chip valid / invalid attribute is minus, such as no device chip (-1), could not be measured by image processing (-2), extracted the device chip (-3), etc. Refer to the center coordinates of the device chip Do not. 0 or plus is used to distinguish the reference measurement point from the calculated coordinate value.

If the reference measurement point is not measured, the coefficients b _{1 to} b ₈ in the equations (1) and (2) cannot be obtained. Therefore, an alternative reference measurement point is set, and that point is added to the reference measurement point list data. The reference measurement point list data has already been copied as work data. This step is step S36. An alternative reference measurement point setting method will be described with reference to FIG. In FIG. 26, a square indicates a device chip, and a thick square indicates a reference measurement point. The device chip in the figure is made up of four regions, each of which has region number (1, 1), region number (1, 2), region number (2, 1) and region number (2, 2). Suppose that

  If the central reference measurement point is not measured, four points (dashed square) around the reference measurement point are set as alternative reference measurement points. The reference measurement points that cannot be measured are referred to the reference measurement point list data to determine which regionized data is related to. That is, it can be seen that the region number (1, 1), the region number (1, 2), the region number (2, 1), and the region number (2, 2). Assuming that there is no device chip at the reference measurement point that cannot be measured, when steps 6, 7, and 8 of the initial measurement region dividing step (FIG. 17) are executed in each region, the broken-line square in FIG. Selected as points and added. If a thicker broken reference measurement point cannot be measured again, a square indicated by a double line is selected and added as a new reference measurement point.

  If there is a verification step, the verification step is executed as step S38, and the subdivision step is executed as step S39. The verification step and the subdivision step constitute a verification process, which is a flowchart shown in FIG. The alignment with the verification step is used to obtain the final version of the regionalized data and the reference measurement point list data, and is used to check the number of the first WLCSP when processing a new variety. It is also used when it is desired to confirm the center coordinates of all device chips immediately after a continuous failure.

  After performing the verification step and the subdivision step in the alignment process of the separated WLCSP, the “finalized version” regionized data and reference measurement point list data obtained by these procedures are stored in the nonvolatile storage device. Save (save final version). “Final version” means that it has been tested. The area data and the reference measurement point list data are statistically determined by statistically examining the difference between the calculated center coordinates of the device chip and the measured center coordinates of several WLCSPs or lots. It is desirable that the error can be determined to be within the allowable value even within the range of variation.

  The method of measuring the reference measurement point is one method for shortening the alignment time. Instead of measuring all the device chip positions on the WLCSP, only the center coordinates of the device chip at the position of the reference measurement point are measured. The center coordinates of each device chip are calculated from the center coordinates of the device chip at the measured reference measurement point. Therefore, when a large number of WLCSPs are processed, the calculated device chip center coordinates accurately approximate the actual device chip center coordinates from the device chip center position of the reference measurement point "stored as a definitive version". Appears when not.

  In the situation as described above, a failure occurs continuously in the testing process or almost the same index position fails in some WLCSPs. However, since it is a statistical phenomenon, it may appear as another situation. In such a case, it is important to first confirm the center position of the failed device chip (confirm whether the calculated center coordinates of the device chip match the center of the failed device chip). . When such a situation occurs in the testing process, as an option, it may be possible to activate processing with a verification step according to the flowchart shown in FIG.

2. Alignment of Semiconductor Device Chips with Device Chip Arrangement After Picker Process As already described, there are two timings for retesting, before picking and after picking. After picking, the device chip array is in a so-called tooth-missing state. The device chip to be picked is managed by the index map shown in FIG. That is, the device chip to be picked must be managed so that the device chip at the arrangement position is extracted in the index map and does not exist (the attribute indicating the validity / invalidity of the device chip in the measurement alignment data − 3). There are many problems in aligning the arrangement of device chips that are in a missing tooth state after being picked.

  As an example of the problem, there is a problem of how to confirm the device chip arrangement on the index map. In FIG. 24, it has been described that a specific position on the device chip arrangement is designated and a corresponding device chip is searched to confirm its existence. However, when the arrangement of the missing device chips is in a complicated state, it is not easy to examine this. A new coordinate reference point and reference direction are introduced to easily align the device chip array in the tooth-missing state after picking in such a situation. The new coordinate reference point and reference direction are based on the dicing frame. The reason for using the dicing frame as a reference is to refer to the alignment data at the first testing.

  WLCSP affixed to the dicing frame is tested several times instead of just once. When WLCSP affixed to a dicing frame is loaded into a CSP probe (or “CSP handler”), there is a possibility that the device chip interval of the separated device chips may change, and WLCSP There is a possibility that the dicing tape attached with will expand and contract. However, according to the present invention, the change in the device chip interval that occurs when loaded into the CSP probe is smaller than the change in the device chip interval that occurs when singulated, and the load is loaded into the CSP probe. Two preconditions are taken: dicing tape expansion and contraction that occurs when it is applied. Here, the expansion and contraction of the dicing tape is small, which means that the device chip on the WLCSP does not deviate more than 1/2 device size (in the X direction and Y direction) from the reference coordinate system of the dicing frame. The effect of dicing tape expansion and contraction is that the same WLCSP is loaded into the CSP probe several times, and the position of the same device chip on the index map is within 1/2 device size in the X and Y directions respectively. It is.

  In addition to using the dicing frame as a reference and referring to the alignment data at the first test, in order to perform alignment in the missing tooth state after picking, it is necessary to newly set a reference measurement point Necessary. Since the tooth is missing, there is a possibility that a reference measurement point in a certain area has already been removed. Check the shape of the measurement area and the reference measurement point for each measurement area. This is performed in another reference measurement point selection step.

2-1. Coordinate management of reference measurement point <Coordinate management step of reference measurement point>
A reference measurement point for managing the center coordinates of the device chip measured as a reference measurement point based on the position arbitrarily set on the dicing frame that holds the dicing tape on which the device chip is placed as a reference position The coordinate management step will be described.

  When aligning a missing WLCSP after picking, it is not desirable to change the algorithm depending on how much the device chip has been removed. Therefore, a method for aligning the missing WLCSP after picking with reference to the alignment data at the first testing is devised. For this purpose, a new coordinate reference point and reference direction are introduced. The shape feature point (reference point) of the dicing frame and the direction of the dicing frame (reference direction) are introduced to manage the coordinates of the reference measurement point.

  The above will be described with reference to FIG. A characteristic portion 56 of the dicing frame 50 is selected as a shape characteristic point (reference point) of the dicing frame. This part 56 is formed with a constriction for determining the position when the dicing frame is fixed to the dicer, and can be easily distinguished from other parts.

  In FIG. 27, a wafer 54 on which WLCSP is formed is affixed to a dicing tape 52 and fixed to a dicing frame 50. A figure indicated by a square indicates a device chip, and a device chip indicated by a bold square in the figure indicates a device chip that provides a reference measurement point for area formation. The coordinates indicated as (X, Y) are an example of the center coordinates of the device chip that gives the reference measurement point for the region. That is, the center coordinates of the device chip at the reference measurement point. In FIG. 27, the dicing frame direction (reference direction) passes through the shape feature point (reference point) 56 of the dicing frame and another shape feature point 57, and an arrow with a rightward direction in the positive direction is attached. Shown as a straight line.

  If the reference point and reference direction of the dicing frame are adjusted to be the same each time, the deviation and WLCSP that occur when the WLCSP is loaded into the measurement table of the CSP probe (or "CSP handler") are pasted to the dicing frame. The effect of time shift is eliminated. If WLCSP is tested several times instead of just once, the device chip spacing of the separated device chips may change when loaded and unloaded to the CSP probe. In addition, there is a possibility that the dicing tape with WLCSP attached will expand and contract. However, here, the condition that such an influence is small when the WLCSP is loaded / unloaded to / from the CSP provider is set. Then, with reference to the alignment data at the time of the first test, the WLCSP in the missing tooth state after picking is aligned.

  There are two reference management point coordinate management steps: a procedure performed at the time of the first test and a procedure performed at the time of the retesting. In the alignment of the separated WLCSP of the first testing, two shape feature points of the dicing frame are measured as the last process, and are stored as measurement data of the WLCSP in the nonvolatile storage device together with the alignment result. The data to be saved will be described later. In alignment at the time of retesting, the measurement data of the WLCSP stored in the nonvolatile storage device is loaded. Then, two shape feature points of the dicing frame are measured, the difference between the angle and the offset from the initial alignment is obtained, and the angle is corrected by rotating the measurement table. The offset is used when calculating the position of the device chip.

2-2. Alignment data for performing retesting In order to perform retesting easily, the characteristic point position (reference point) of the dicing frame and the direction of the dicing frame (reference direction) are determined at the first test. Need to be measured during alignment. The feature point position (reference point) and dicing frame direction (reference direction) of the measured dicing frame, together with the alignment result data (measurement alignment data), name the frame reference alignment data,
Frame reference alignment data header part Number of arrays (X direction, Y direction) Kx, Ky
Reference point X _f , Y _{f of} dicing frame
Dicing frame direction (reference direction) T _f
Data part of frame reference alignment data Attribute id [i] [j] indicating device chip valid / invalid
Center coordinates of device chip (px [i] [j], py [i] [j])
Are stored as WLCSP measurement data in a non-volatile storage device (such as a hard disk). The WLCSP measurement data is workpiece measurement data, and one WLCSP is one measurement data. This is not necessary if you are not retesting.

  As the last process (process to be placed after step S35) described in the procedure for performing alignment with the separated WLCSP (process placed after step S35), the feature point position (reference point) of the dicing frame and the direction of the dicing frame Measure (reference direction). The characteristic point position (reference point) of the dicing frame is measured by measuring the positions of two notches in the dicing frame by image processing. For this, a normal pattern matching method can be used.

2-3. Alignment for retesting After picking, the arrangement of device chips is missing. The area boundary points (separate reference measurement points) are arranged, and the device chip array (index map) on the WLCSP is divided into measurement areas. You cannot use regionalized data. In some measurement areas, the reference measurement point may have already been removed. It is necessary to check the shape of the measurement area and the reference measurement points for each measurement area, and to make up for any shortages. This is performed in another reference measurement point selection step.

<Another reference measurement point selection step>
The reference measurement point needs to be reset for retesting. Do not rewrite the regionalization data and reference measurement point list data stored in the storage device as part of the parameter set of the product parameters already described, but re-write them as data used only for retesting temporarily. Set. This resetting step is the content of another reference measurement point selection step.

  With reference to FIG. 28, the contents of the above-described another reference measurement point selection step will be described. The portion where the device chip is picked is shown by a small square with a broken line. Here, description of portions common to FIG. 27 is omitted. A region indicating a region 58 for resetting the reference measurement point is indicated by a broken-line rectangle. As shown by a small square with a broken line at the position indicated by (X ′, Y ′) in FIG. 28, there is a reference measurement point extracted by picking after the first test.

  The different reference measurement point selection step includes the following three processes. 1). Check whether the reference measurement point of each area is picked. 2). Select at least 3 reference measurement points in each area. 3). For each region, step 6, step 7, and step 8 of the method described in section 1-2 “Division method of region” are performed.

  The area is the same as the stored area data. The area is not newly set again. The index position of the reference measurement point corresponding to the in-area reference measurement point number is obtained from the reference measurement point list data for each region. When the device chip at the index position is picked, the reference measurement point cannot be used. In this way, it is examined how many of the reference measurement points can be used. In an area where three or more reference measurement points cannot be secured, a reference measurement point is added by the following method. First, the number of device chips belonging to the area is checked (because device chips are extracted by picking).

When the number of device chips is 3 or less, the inspection of the area is completed using all device chip positions (index positions) as reference measurement points. If the number of device chips is 4 or more, the device chips located at the upper left, lower left, upper right, or lower right points among the device chips belonging to the area are the three points that are not reference measurement points in order. Select until (upper left, lower left, upper right, lower right). The method of selecting the upper left, lower left, upper right, or lower right point is (i _X + i _Y ) for all device chip positions (index positions are i _X , i _Y ) that are not extracted. (i _Y −i _X ) can be calculated and (i _X + i _Y ) and (i _Y −i _X ) can be obtained as index positions where the maximum and minimum values are obtained (this method is the same as described in FIG. 24). The same way).

  Since the position and range of the region to be checked can be easily determined from the start index number and the region index number, it is not difficult to check all device chips belonging to the region. After the determination of three or more reference measurement points, the procedure 6, procedure 7 and procedure 8 described in section 1-2 “Division method of region” are performed, and the reference measurement point of that region is obtained. When the process is completed for all the areas, the separate reference measurement point selection step ends.

When the different reference measurement point selection step is completed, preparation for alignment (retesting alignment) with respect to the device chip array in the missing tooth state after picking is completed. Organizing the data needed for retesting alignment,
1). Regionalized data re-set in another reference measurement point selection step 2). Reference measurement point list data re-set in the separate reference measurement point selection step 3). It becomes three pieces of frame reference alignment data at the time of the first testing. The area data of 1) and the reference measurement point list data of 2) are not stored as product type data, but are for a device chip array in a missing tooth state to be retested.

  FIG. 29 is a flowchart showing an alignment procedure in retesting. Data necessary for alignment of the WLCSP to be retested is loaded from the nonvolatile storage device (step S45). There are three types of data to be loaded: area data as model data, reference measurement point list data, and frame reference alignment data at the first test. When the data is loaded, the characteristic point position (reference point) of the dicing frame and the direction of the dicing frame (reference direction) are first measured (step S46: Dicing frame measurement (angle adjustment)). The angle adjustment is performed so that the angle measured as the dicing frame direction is the same as the angle of the frame reference alignment data. When the angle is adjusted in this way, the angle is adjusted to the same angle as that in the initial testing (coordinate management step of the reference measurement point).

  Next, another reference measurement point selection step is executed (step S47). In the separate reference measurement point selection step, the area data and the reference measurement point list data are reset. If picking is not performed, the area data and the reference measurement point list data are the same as those loaded (since there is no change in the index map). The WLCSP alignment is executed with reference to the three data of the reset regionized data, the reset reference measurement point list data, and the frame reference alignment data at the first test. WLCSP alignment is performed by sequentially measuring the center coordinates of the device chip located at all the reset reference measurement points by image processing using pattern matching (step S48). It is checked whether or not the center coordinates of the device chip located at the reference measurement point have been measured (step S49). A reference measurement point for which the center coordinate of the device chip is not measured belongs to a certain region in which an alternative reference measurement point is set and is added to the reconfigured regionized data and the reconfigured reference measurement point list data (step S51). When the center coordinates of the device chips located at all the reference measurement points are measured, the device chip center coordinates of the remaining unmeasured index positions can be obtained by calculation. In this way, when the center coordinates of the device chip located at the reference measurement points in all the areas are measured, all the device chip center coordinates on the WLCSP are obtained (step S50). The alignment ends.

  In the retesting alignment, the position of the dicing frame is adjusted to be the same as that in the initial testing. If the position of each device chip on the WLCSP is not displaced between the initial testing and the retesting, the frame reference alignment data at the initial testing can be used as it is for the retesting. However, by the time of retesting, there are several loading / unloading effects on the CSP provider (or “CSP handler”), and there is also the effect of picking. These influences are expansion and contraction of the dicing tape and changes in the device chip interval. In the present invention, an alignment method was devised on the premise that these effects are small.

  When the center coordinates of the device chip located at the reset reference measurement point are measured by image processing, the measured coordinates are slightly different from the coordinates of the frame reference alignment data at the first testing. This difference is due to several loads and unloads to the CSP provider and picking. Correcting this difference is the purpose of retesting alignment. Since the precondition that this difference is small is provided, when the center coordinates of the device chip located at the reset reference measurement point are measured by image processing, the designated position does not change greatly. That is, in image processing, a position closest to a designated position may be used as a measurement value.

  An option for retesting alignment is the process of adding a test step. In the verification step, the center coordinates of all device chips on the WLCSP are measured, the coordinates calculated from the reference measurement points are compared with the measured coordinates, and if the difference in coordinates is large, the measurement area is divided (subdivision) Step). However, the retesting alignment is basically performed only once. The device chip arrangement in the missing state after picking becomes a different missing state each time. For this reason, the regionized data and reference measurement point list data that are reset in the separate reference measurement point selection step are temporarily used only for retesting. Since the test result of data that is used only once is meaningless, performing the test step is synonymous with actually measuring the center coordinates of all device chips.

  When retesting is performed before picking, it may be performed as an alignment procedure of WLCSP separated into pieces (alignment procedure that is not in a missing state). In this case, the alignment process does not require distinction between initial testing and retesting.

3. Deviation adjustment method when testing individualized device chips A measurement electrode (installed in a prober) connected to a device chip tester for evaluating the electrical characteristics of the device chip and a device electrode of the device chip Consider the positional deviation (also referred to as the difference) allowed in testing when the work to determine the electrode position of the device chip is performed to align the positions (hereinafter referred to as the measurement electrodes connected to the device chip tester) Is sometimes referred to as the “device chip tester measurement electrode”). The deviation between the measurement electrode of the device chip tester and the device electrode of the device chip required for evaluating the electrical characteristics of the device chip, that is, the allowable deviation is the so-called necessary degree of matching.

  With reference to FIG. 30, the status of the testing process in multi-site testing will be described. In FIG. 30, the horizontal axis represents the X axis and the vertical axis represents the Y axis, and the presence of the device chip is indicated by a square indicated by a solid line. In addition, a thin broken-line square indicates an index position in this drawing, but indicates a position where no device chip actually exists.

  In general, in testing, as shown in FIG. 30, the device chip on the measurement table is relatively moved so that the measurement electrodes of the device chip tester move to the required positions as shown in FIG. The position is moved to and the measurement is performed. Hereinafter, for simplicity, the device chip tester is sometimes simply referred to as a tester.

  The center position of the measurement electrode of the tester is fixed for each device chip. On the other hand, the center position of the device chip side terminal is slightly deviated from the design value of the device chip interval because the device chip is separated. For this reason, in actual testing, it is necessary to perform the correction described with reference to FIGS. 6 and 7 every time a test is performed.

  With reference to FIGS. 31A and 31B, how to correct the above-described deviation in measurement will be described. FIG. 31 (A) shows the measurement electrodes of the tester, and FIG. 31 (B) shows the device chip side terminals, respectively, with the X axis on the horizontal axis and the Y axis on the vertical axis. In each figure, the four large rectangles correspond to the device chip, the small circles represent the tester measurement electrodes (pogo pins 92), and the small rectangles represent the device chip terminals (pumps). Terminal 94) is shown. 31A and 31B, the center position of the measurement electrode of the tester in the device chip unit and the center position of the device chip side terminal in the device chip unit are indicated by + marks.

  The center position of the measurement electrode of the tester is fixed, and its coordinate value has a fixed value. On the other hand, the individual device chip center coordinates on the device chip side are calculated from the device chip center coordinates of the measured reference measurement points.

  As shown in FIG. 30, the testing seems to be performed while moving the index map in order of step 1 and step 2. In FIG. 30, since 4 (2 × 2) multi-testing is described as an example, one region (range enclosed by a broken line or a solid square) in which two device chips are arranged vertically and horizontally is described. Testing is performed as a unit.

  Referring to FIGS. 32 (A) and (B), the center position of the measurement electrode of the tester (the position indicated by the solid + mark in FIG. 32 (A)) and the center position of the device side terminal (FIG. 32 (B The relationship between the positions indicated by a broken line +) will be described. When the individual device chip center coordinates are calculated from the device chip center coordinates of the reference measurement point, the device chip center position is obtained as indicated by the broken + mark in FIG. 32 (B). The center position of the measurement electrode of the tester in FIG. 32 (A) and the center position of the device chip side terminal in FIG. 32 (B) are matched by performing parallel movement and rotational movement with respect to the device chip. A coordinate transformation representing this movement is given by the following equation.

X = x × cosθ + y × sinθ + dx ₀ (3)
Y = −x × sinθ + y × cosθ + dy ₀ (4)
Here, θ is a rotation angle, and dx ₀ and dy ₀ are parallel movement amounts in the X-axis direction and the Y-axis direction, respectively. Further, (x, y) is the center coordinate of the device chip, and (X, Y) is the center coordinate of the measurement electrode.

In multi-site testing, the number of device chips to be measured simultaneously is two or more, while the equations (3) and (4) include three parameters, θ, dx _0, and dy ₀ as parameters. Therefore, the least square method is used to obtain these parameters.

From the relationship between the center position of the measurement electrode of the tester and the center position of the device chip side terminal shown in FIGS. 32 (A) and (B), the parameters of Equations (3) and (4), θ, dx ₀ and When dy ₀ is obtained and superimposed using the calculation result of the center position of the measurement electrode of the tester and the center position of the device chip side terminal, the overlapping state is as shown in FIG. FIG. 33 shows a state where the center position of the measurement electrode and the center position of the device chip are best overlapped by performing a parallel movement and a rotational movement with the device chips of the entire group to be multitested as one lump. It is a figure.

Substituting the obtained parameters, θ, dx ₀ , dy ₀ and the center coordinates of the device chip into the equations (3) and (4), the center coordinates of the device chip after movement (translation and rotation) are obtained. The difference (deviation) between the center coordinates of the measurement electrode of the tester and the center coordinates of the device chip is obtained. This deviation must be less than an allowable value that does not damage the device chip side electrode and does not affect the measurement. That is, it must be within the required degree of match.

  In the example shown in FIG. 33, the difference between the center positions of the measurement electrode of the tester and the device chip side terminal at the position denoted by 2 is large. A device chip position where the difference exceeds an allowable value is treated as being untestable. Since the device chip position is not tested, the device chip position is deleted (as a device chip that cannot be tested), and the above-described process by coordinate transformation is performed again (to make the remaining device chips overlap at better positions). Device chip positions that differ by more than the allowable value cannot basically be aligned with the measurement electrodes of the tester in multi-site testing. In such a situation, an error may occur in the CSP prober (or “CSP handler”).

  The example shown in FIG. 33 is an example of 4 (2 × 2) multi-testing. When the number of multis is large, the difference between the center position of the measurement electrode of the tester and the center position of the device chip can be further varied. Situation arises. If the difference exceeds the allowable value at 2 points or more (for example, the position indicated as 2 and the position indicated as 4 in FIG. 33), that is, if the difference is not within the required degree of matching, the maximum difference will occur. The data for the current position is deleted from the process and the process is repeated. The points with the largest difference are deleted sequentially so that the remaining points are within the allowable value.

  Performing rotation correction by testing generally results in poor throughput. Therefore, an option not to perform rotation correction is prepared for Equations (3) and (4). There are two options for not performing rotation correction.

  One is that no rotation correction is performed. This is equivalent to the equation (3) and (4) where θ = 0. In the equations (3) and (4), if θ = 0, the following equation is obtained.

X = x + dx ₁ (5)
Y = y + dy ₁ (6)
For the option of not performing any rotation correction, the same method may be executed by using equations (5) and (6) instead of equations (3) and (4).

In another option, when the correction angle is small enough not to perform the rotation correction, the rotation correction is not performed. For this purpose, the previous testing angle (θ ₀ ) is fixed and only the translation component is obtained. That is, in the following formulas (7) and (8), θ ₀ is a known value and only the parameters dx ₂ and dy ₂ are obtained by the least square method. Therefore, the conversion formula in this case is X = x × cos θ ₀ + y × sin θ ₀ + dx ₂ (7)
Y =-x x sin θ ₀ + y x cos θ ₀ + dy ₂ (8)
become that way.

  With reference to FIGS. 34A and 34B, the relationship between the center position of the measurement electrode of the tester and the center position of the device chip side terminal when the rotation correction is not performed and when the rotation correction is performed will be described. The center position of the measurement electrode of the tester is indicated by a solid + mark, while the center position of the device side terminal is indicated by a broken + mark. Of course, the difference is smaller when the rotation correction shown in FIG. 34 (B) is performed, but even if the rotation correction shown in FIG. 34 (A) is not performed, if the difference is less than the allowable value, testing can be performed. .

The rotation correction in the testing is performed with respect to the rotation center position of the measurement table. Therefore, assuming that the rotation center position (known) of the measurement table is (x _c , y _c ), the equations (3) and (4) are changed to the following equations (9) and (10).

X = (x − x _c ) × cosθ + (y − y _c ) × sinθ + dx ₃ (9)
_{Y = - (x - x c} ) × sinθ + (y - y c) × cosθ + dy 3 (10)
The parameters dx ₃ and dy ₃ obtained from the equations (9) and (10) are obtained from the parameters dx ₀ and dy ₀ obtained from the equations (3) and (4). The parameters representing the parallel movement with respect to the rotation center position of the measurement table are dx ₃ and dy ₃ . The coordinates of individual device chips measured by alignment are obtained based on the angle adjusted in the first process of the WLCSP alignment flow shown in FIG.

4). Deviation adjustment method and alignment procedure when the device chip is large If the device chip size is large, the displacement at the time of singulation is small, but the tilt of the device chip is different from the case when the device chip size is small. appear. In order to examine this problem, the overlapping state of the measurement electrode of the tester and the device chip side terminal with respect to the inclination of the device chip will be described with reference to FIGS. 35 (A) and (B). FIG. 35 (A) shows the case where the device chip size is small, and FIG. 35 (B) shows the degree of overlap between the measurement electrode of the tester and the device chip side terminal when the device chip size is large. In FIGS. 35A and 35B, large squares (104a and 104b) indicate the outer periphery of the device chip, and white circles (106a and 106b) indicated by broken lines indicate tester measurement electrodes (pogo pins). The small squares (108a and 108b) that overlap with each other and look black are device chip side terminals (bumps).

  As described above with reference to FIG. 5, the measurement electrode of the tester is fixed so that a plurality of device chips can be measured simultaneously, as described above with reference to FIG. The angle cannot be adjusted. On the other hand, since the device chips are separated into individual pieces, the inclination angle of each device chip may change independently.

  As shown in FIG. 35 (B), when the device chip size is increased, there is an electrode that is largely displaced between the measurement electrode and the device chip side terminal. With respect to the tilt of the device chip, even an angle that does not cause a problem in a small-sized device chip is a problem in a large-sized device chip.

  Until now, the alignment angle of the device chip has not been measured in the alignment. However, in order to cope with the situation as described above, it is necessary to measure the device chip center position and the tilt angle of the device chip with respect to all device chips on the WLCSP.

  In order to measure the center position and inclination angle of all device chips using image processing techniques such as pattern matching, it is necessary to use two model image patterns instead of one. As an alignment result, both position data (for example, device chip center coordinates) and an angle must be held. When the device chip size is large, the number of device chips is reduced, but it may be necessary to measure the center coordinates and inclination angles of all device chips by alignment. In addition, it may be difficult to perform multi-site testing under the situation where the tilt of the device chip is a problem. When the angle between a plurality of device chips increases with a chip having a large device chip size, it becomes difficult to simultaneously overlap the terminals of the plurality of device chips with the measurement electrodes on the tester side. In this case, single site testing must be performed.

FIG. 36 shows the relationship between the individual positions and the center positions of the tester measurement electrodes in the two multi-testing. Assume that the position of each measurement electrode is known with respect to the center position of the tester measurement electrode, and the relative coordinates of the jth electrode are (X _pj , Y _pj ). That is, each position of the tester measurement electrode is obtained as a design value, and is handled as known with respect to the center position of the tester measurement electrode.

The center position and inclination of the tester measurement electrode must be measured separately. If the center coordinates of the tester measurement electrode are (X _n , Y _n ) and the inclination of the tester measurement electrode is θ _0n , the position (X _nj , Y _nj ) of each measurement electrode is calculated according to the following equation.

X _nj = X _n + X _pj × cosθ _0n + Y _pj × sinθ _0n (11)
Y _nj = Y _n − X _pj × sinθ _0n + Y _pj × cosθ _0n (12)
The subscript n is a number for identifying the measurement electrode for each device chip. Here, (X _n , Y _n ) and θ _0n are measured values, and (X _pj , Y _pj ) are design values, both of which are known. That is, when (X _n , Y _n ) and θ _0n are measured, the coordinates of each measurement electrode are known. Further, when the coordinates of each measurement electrode can be directly measured, the measurement coordinates (X _nj , Y _nj ) obtained by direct measurement are used without using the equations (11) and (12).

Next, the position of the terminal on the device side will be examined in the same manner. FIG. 37 illustrates the relationship between the center position of a device chip and individual device chip terminals in two multi-testing. Let (x _qj , y _qj ) be the relative coordinates from the center position of the device chip for each terminal of the device chip. As the device chip position, center coordinates and an inclination angle are measured. Assuming that the measured device chip center coordinates are (x _n , y _n ) and the inclination angle with respect to the x axis is θ _1n , each device chip terminal coordinate (x _nj , y _nj ) is expressed by the following equation (13): And (14).

x _nj = x _n + x _qj × cosθ _1n + y _qj × sinθ _1n (13)
y _nj = y _n − x _qj × sinθ _1n + y _qj × cosθ _1n (14)
The subscript n is a number for identifying each device chip, and is the same as the case of the measurement electrode described above. The relative coordinates for the center of the j-th electrode are (X _pj , Y _pj ) and the relative coordinates for the center of the j-th terminal are (x _qj , y _qj ), but basically X _pj = x _qj , Y _pj = y _{There is a qj} relationship.

  Thus, the coordinates of each of the tester measurement electrodes and the coordinates of each of the device chip terminals are represented, and the expressions (3) and (4) are changed to the following expressions (15) and (16). n is a number for identifying each device chip unit, and j is a number for indicating the order of electrodes or terminals.

X _nj = x _nj × cosθ + y _nj × sinθ + dx (15)
Y _nj = −x _nj × sinθ + y _nj × cosθ + dy (16)
The center coordinates (X _n , Y _n ) of the tester measurement electrode and the inclination θ _0n of the tester measurement electrode, and the center coordinates (x _n , y _n ) of the device chip and the inclination angle of the device chip are set to θ _1n Substituting into the equation (16), the parameters θ, dx and dy are obtained by applying the least square method. The obtained parameters θ, dx, and dy give the amount of translation and the rotation angle at which the corresponding tester measurement electrode and device chip terminal most closely overlap. Further, using the obtained parallel movement amount and rotation angle, when the difference between Equation (15) and Equation (16) is taken, Equation (17) representing the deviation (difference) between each measurement electrode and the device chip terminal And (18).

δx _nj = X _nj − (x _nj × cosθ + y _nj × sinθ + dx) (17)
δy _nj = Y _nj − (−x _nj × sinθ + y _nj × cosθ + dy) (18)
The deviation between each measurement electrode and the device chip terminal must be less than the allowable value. That is, it must be within the required degree of match.

  In the case where the device chip size is large, the method of adjusting the tester measurement electrode and the device chip terminal so that the deviation between the device chip terminals becomes small has been described. In such a case, the center of all device chips is described. It is necessary to measure coordinates and tilt angles. In single-site testing, the center coordinates and tilt angles of all device chips should be measured. The alignment process for measuring the center coordinates and tilt angles of all device chips is shown in the flowchart of alignment of individual WLCSPs shown in FIG. 22, and after the correspondence between the index map and each device chip is taken, Measure the center coordinates and tilt angles of all device chips. In order to measure the center coordinates and tilt angle, it is necessary to measure two or more different positions with a single device chip using the pattern matching method.

5. Alignment procedure when device chip is small The description of alignment up to the previous section 4 has assumed that the size of the device chip to be tested is 2 mm x 2 mm or more. No problems have been described so far when the device chip size is smaller than this.

  In addition, from the viewpoint that device chips can be transported by dicers currently in circulation, the maximum allowable amount of dicing tape is about 5 mm to 6 mm. The explanation is made assuming that the distance is 1 mm or less.

  If the size of the device chip is 2 mm × 2 mm or more, the situation as shown in FIG. 4, that is, the Y-direction center line of the device chip located near the center of the wafer is one lower toward the end of the WLCSP. The situation where the position of the row device chip is reached has not occurred. Although the maximum extension of the dicing tape is estimated to be 0.5 mm or less in practice, it has been ensured that the implementation of the present invention can be performed with an allowance by assuming this to be a little larger than 1 mm.

  The maximum extension amount of the dicing tape depends on various factors such as the type of tape, the condition of singulation, and the device chip size. As the device chip size decreases, the number of cuts in the dicing tape increases, and the amount of expansion of the dicing tape increases. Also, the force with which each device chip opposes the dicing blade is weakened as the device chip size is reduced.

  For this reason, as the device chip size becomes smaller, the device chip position is greatly displaced. When the device chip size is small, it is necessary to perform alignment in consideration of such influence. In the descriptions that have been made so far, as a guideline to consider such a situation, it has been assumed that a device chip size of a small size is 2 mm × 2 mm or less.

  With reference to FIG. 38, the above will be specifically described for an example of a WLCSP having a small device chip size. When the device chip size is small, as already described with reference to FIG. 4, the Y-direction center line of the device chip existing near the center of the dicing tape is one row device chip below the end of the dicing tape. (Or the position of the row device chip one level above). The same thing happens in the column direction as in the row direction.

In order to perform multi-site testing, device chips must be arranged in a two-dimensional array arranged to some extent even with small device chips. In order to avoid the possibility of being indistinguishable from the position of the lower (or upper) row device chip, given the flow of the separated WLCSP alignment shown in FIG.
1) A process for adjusting the angle of the WLCSP placed on the measurement table 2) A process for setting a reference measurement point 3) An additional means described below is added to the three processes of taking the correspondence between the index map and each device chip Add.

  For the process of adjusting the angle of the WLCSP placed on the measurement table, in the process described with reference to FIG. 23, the vicinity of the center (first point), the left of several device chips (second point), the left end A total of five points (the third point) and the right end (the fourth point) were searched by image processing. For a small device chip, the process shown in FIG. 23 is modified to the process shown in FIG. When searching from the second point to the third point (all when searching for the next point after the third point), in FIG. 39, the search is performed so that the number of device chips is five as shown in the figure. The device chip position is determined. When a small device chip is singulated, the device chip interval expands and contracts (generally expands). Therefore, when a certain number of device chips are separated, they cannot be distinguished from the adjacent device chips. In the angle adjustment process, it is necessary to determine the maximum number of device chips for angle adjustment and search at positions equal to or less than that number (cannot be regarded as the same row). The maximum number of device chips for angle adjustment will be described later.

  FIG. 39 is a diagram for explaining a measurement position for adjusting the angle of the WLCSP placed on the measurement table. FIG. 39 shows a row of device chips arranged in a line, and a small square portion indicates a device chip. In the case described with reference to FIG. 39, the angle adjustment maximum device chip number is set to five. In the case described with reference to FIG. 23, when determining the fourth point search position, the line connecting the first point, the second point, and the third point is determined by the least square method, and the straight line is extended. The line was set to be near the left edge of WLCSP. In the case described with reference to FIG. 39, the number of operations for setting the next search position from the already searched position increases. In such a case, the search position is set on the extension line of the two obtained points. It will be set. On the other hand, for example, when the sixth point search position shown in FIG. 39 is set, it is assumed that the position exists on the extension lines of the first point and the third point. In the case described with reference to FIG. 39, a line connecting the eight points measured in this way is calculated by the least square method, an angle is obtained from the calculated slope of the line, and this angle is calculated as the slope of WLCSP. As a correction. If the device chip size is smaller, more points may be searched and the angle of the WLCSP placed on the measurement table may be adjusted.

  The maximum search device chip number and the maximum angle adjustment device chip number will be described. The maximum search device chip number is the maximum number that does not change the position corresponding to one device chip interval when searching for device chips in the arrangement direction of the device chips. When the device chips are separated, the device chip interval expands and contracts (generally expands).

Elongation rate of device chip interval = (maximum device chip interval / design value of device chip interval)-1.0
It is defined as One example is taken with the device chip spacing design value of 1.0 and the device chip spacing elongation of 10%. 1.0 / Elongation rate of device chip interval = 10 or more away, the number of device chips in between becomes not unique. If there is a device chip at a position 12.05 away from a certain device chip, it is not known whether there are 11 chips at intervals of 1.004 or 10 chips at intervals of 1.095. The maximum number of search device chips is determined from the growth rate of the device chip interval as follows (-1 is a margin).

Maximum number of device chips searched = 1.0 / Device chip interval growth rate-1
The device chip size may be different in the X and Y directions, and the influence when singulating may be different in the X and Y directions, so the maximum search device chip number is set separately for the X and Y directions. I will be able to do it.

  The angle adjustment maximum device chip number is a maximum number that is not mistaken for the device chip in the upper or lower row in the angle adjustment process. The maximum device chip number in the X direction angle adjustment is related to the elongation rate of the Y direction device chip interval. To adjust the angle, the device chip cannot be discriminated when it is displaced by 50%.

X direction angle adjustment maximum number of device chips = 0.5 / Elongation rate of Y direction device chip interval -1
The maximum Y-direction angle adjustment device chip number is determined in the same manner. In the angle adjustment process shown in FIG. 39, the device chip interval is equal to or less than the maximum number of device chips for angle adjustment.

The process of setting the reference measurement point is performed in the initial measurement area dividing step. In the initial measurement area dividing step, it is necessary to set an initial value of the number of area indexes. When the initial value of the area index is set, in the initial measurement area dividing step, the device array on the WLCSP is divided into measurement areas as shown in FIG. 18, and the reference measurement points are also set at the same time. When the device chip size is small, the initial value of the number of area indexes is set to the above-described maximum number of search device chips when the largest value is set. Normally, about 1/2 of the maximum number of search device chips is preferably set as the initial value of the number of area indexes. The search maximum device chip number is the maximum number that does not make a mistake in one device, but the region index number is a device chip number that does not greatly change the device chip interval within that range. Naturally, the maximum number of search device chips> the number of regional indexes. If the initial value of the number of regionalized indexes is too large, it is necessary to perform an examination step and a re-division step on a large number of WLCSPs after the initial measurement region division step. When the device chip size is small, it is necessary to perform the initial measurement region dividing step in consideration of this point. When the device chip size is small, one additional means is implemented in the initial measurement area dividing step, which will be described later (the same as the contents described in the process of taking the correspondence between the index map and each device chip). ).

  The process of taking the correspondence between the index map and each device chip for a large device chip of 2 mm × 2 mm or more can be easily executed as described with reference to FIG. This is because a situation in which the position cannot be distinguished from the position of the lower (or upper) row device chip does not occur. However, in the case of a small device chip, it is necessary to use many device chip positions at the same time in order to make the correspondence between the index map and each device chip.

  A method of taking correspondence between the index map and each device chip will be described with reference to FIG. In FIG. 40, a thin line square indicates a device chip, and a thick line square indicates a device chip of a reference measurement point. With a small device chip size, when searching for the position of one device chip and then searching for the next device chip position, the principle is that the maximum number of device chips to be searched must not be exceeded. Therefore, when measuring the device chip center coordinates of the reference measurement point shown in FIG. 40, the measurement order is determined so that this principle is observed. Draw a 45-degree slant line to the right, and then move this straight line from top to bottom in the Y direction to pick up the reference measurement points located on the straight line in order, thereby changing the measurement order of the reference measurement points. I will decide. When there are a plurality of reference measurement points on the same straight line, a rule from above is provided. In FIG. 40, the measurement order from 1 to 8 is added to the reference measurement point in this way. The order of measurement can be determined in eight ways by using a 45-degree oblique line, X-direction line, and Y-direction line (another order determination method such as Line 1 'in FIG. 40). ). Even in the case of using a line in the X direction, numbers can be assigned in order (by distance) close to the line passing through the center of the WLCSP.

  When the device chip center coordinates of the reference measurement points are measured in the order of measurement and the device chip center coordinates of all the reference measurement points are measured, the index map can be matched with each device chip and alignment can be performed at the same time. It will be completed. In such a small device chip, the correspondence between the index map and each device chip and the measurement of the reference measurement point need to proceed simultaneously. Therefore, the measurement order of the reference measurement points was decided. In order to improve that the correspondence between the index map and each device chip cannot be taken unless all the reference measurement points are measured, the following correspondence is taken. In FIG. 40, reference measurement points 1 to 8 exist on the upper left side of the WLCSP, and there are positions that do not exist in the adjacent device chip. At reference measurement points 1 to 8 in FIG. 40, the center coordinates of the device chip are measured, and it is further checked whether there is an adjacent device chip. The method of coding the presence / absence of the adjacent device chip is a process of already taking the correspondence between the index map and each device chip, and has been described with reference to FIG. The same method is used here. Therefore, in the case of small device chips, it is necessary to check whether or not there is a device chip adjacent to the reference measurement point in the process of setting the reference measurement point (initial measurement area division step and subdivision step) (reference measurement point list data Adjacent device chip presence / absence data).

  In FIG. 40, when the device chip center coordinates of the reference measurement points 1 to 8 are measured, it is checked whether or not the adjacent device chips are coincident. If the presence or absence of the adjacent device chip matches at the reference measurement points 1 to 8 in FIG. 40, the correspondence between the index map and each device chip can be taken (since it is determined to have been taken), and at the subsequent reference measurement points, the device Measure only the center coordinates of the chip. Reference measurement points must be measured in order. In image processing, even if there is the same pattern, pattern matching may not be possible, so at the reference measurement points from 1 to 8 in FIG. Necessary (judgment conditions may be slightly loose). When the reference measurement points 1 to 8 in FIG. 40 are measured and it is determined that the correspondence between the index map and each device chip cannot be obtained, the measurement order of the reference measurement points is changed. The method for determining the measurement order of the reference measurement points has already been described. When the additional means described above is added to a small device chip, the alignment can be performed by the individualized WLCSP alignment flow shown in FIG.

6). Schematic Configuration Example of Device Chip Position Measurement System With reference to FIGS. 41 and 42, a schematic configuration example of a CSP prober equipped with a device chip position measurement system that implements the device chip position measurement method of the present invention will be described. . Note that the CSP prober described with reference to FIGS. 41 and 42 is merely an example in which the present invention can be suitably used, and the configuration of the CSP prober that can use the present invention is not limited to the configuration shown in this figure. Moreover, each component is only shown to such an extent that the schematic structure of the system for implementing the method of this invention can be understood. The conditions described below are merely preferred examples.

  First, a schematic configuration of a CSP prober will be described with reference to FIG. The CSP prober includes a CSP prober control device 110 and a device chip position control device 150. The CSP prober controller 110 includes a measuring unit 120 and a CSP prober controller 130. The measurement unit 120 includes a measurement electrode 122 and a measurement table 124, and has a structure capable of performing multi-site testing. The measurement electrode 122 is installed in a CSP prober and is directly connected to the device chip tester 200 by a measurement cable (sometimes referred to as “measurement electrode of a device chip tester”). The positional relationship between the measurement electrode 122 and the measurement table 124 is calibrated before the electrical characteristic inspection is performed. In addition, it is confirmed directly by conducting a preliminary test and examining the trace of the measurement electrode remaining on the device chip terminal electrode.

  The CSP prober control device 110 can be configured using, for example, PCP-100SL manufactured by Prime Five Co., Ltd. disclosed in Non-Patent Document 1. Of course, the present invention is not limited to this PCP-100SL, and it can be configured by using a normal prober having a function capable of testing while simultaneously attaching a large number of device chips to a dicing tape.

  The CSP prober control unit 130 gives an instruction for contact with the device electrode of the device chip or detachment from the device electrode to the measurement table 124 that constitutes the measurement unit 120, gives an instruction about movement of the measurement position, etc. Give instructions for measuring device chip electrical characteristics.

  Further, the CSP prober control unit 130 provides information for giving instructions to the measurement table 124 about contact of the device chip to the device electrode, separation from the device electrode, or movement of the measurement position, etc. Has a function to acquire from 150.

  The device chip position control device 150 acquires information such as the positions of device chips that are affixed to the dicing tape in a singulated state from the camera 148, and performs initial measurement area division step, area data load A function for executing processing including a step, an image processing step, a verification step, a subdivision step, a coordinate management step for a reference measurement point, and another reference measurement point selection step. That is, the device chip position measurement system is built in the device chip position control device 150.

  The device chip position control device 150 includes an image processing device 115, a first arithmetic device 160, a first data output unit 162, a first data input unit 164, a display, in order to realize the processing function of the device chip position measurement system described above. A device 166 is included. The image processing apparatus 115 includes an image capturing unit 152, an image conversion processing unit 154, and a first frame memory 156.

  The image capturing unit 152 captures video information of the device chips that are captured and sent by the camera 146 and that is separated into pieces and pasted on a dicing tape and arranged in a matrix, and converts the video information into an image conversion process. Delivered to part 154.

  These image conversion processes executed in the image conversion processing unit 154 can be selected and used from among a number of existing image conversion methods. Since these image conversion processes are well-known matters, description thereof will be omitted (see, for example, JP-A-2001-264037).

  In the image conversion processing unit 154, the image information of the device chip that has undergone the necessary image conversion processing is temporarily stored in the first frame memory 156, and the temporarily stored image information is stored in the first frame memory 156. To the first arithmetic unit 160.

  A schematic configuration example of the first arithmetic unit 160 will be described with reference to FIG. The first arithmetic unit 160 includes a control unit 170, a central arithmetic processing unit 180, and a nonvolatile storage device 172. The control unit 170 controls the central processing unit 180 and the nonvolatile storage device 172 that constitute the device chip position measurement system. The central processing unit 180 includes an initial measurement region dividing unit 182, an image processing unit 184 including a device chip position interpolation unit 184a, a reference measurement point and measurement region management unit 186, a measurement region adaptation unit 188, and a frame reference point management unit. Comprised of 190.

  The nonvolatile storage device 172 includes a part 172a for storing an index map in which each device chip is specified in association with a two-dimensional array (two-dimensional index), a regionized data, a reference measurement point list data, and the like as a set. A part 172b for storing the parameter set and a part 172c for storing frame reference alignment measurement data which is alignment measurement data for use in retesting are included. The nonvolatile storage device 172 is also used for “save candidate” or “save finalized version” in the division step. The nonvolatile storage device 172 can be configured using an existing hard disk or the like.

  The initial measurement area and the initial reference measurement point are set by the initial measurement area dividing unit 182 based on the index map information input in advance to the nonvolatile storage device 172 and the instruction information given from the display device 166. Thus, it is written in the nonvolatile storage device 172 as a parameter set of product parameters.

  The image processing steps executed by the first arithmetic unit 160 are programmed so as to be able to execute the above-described device chip position interpolation calculation process, and are stored in the nonvolatile storage device 172. When the control unit 170 gives the image processing step execution instruction to the image processing unit 184, the measurement area data and the reference measurement point data stored as the parameter set are read from the nonvolatile storage device 172, and these data are read as the reference measurement points. And the measurement area management unit 186. An instruction for moving the measurement table 124 to the device chip position of the reference measurement point managed by the reference measurement point and measurement area management unit 186 is issued by the image processing unit 184 via the CSP prober control unit 130. After the movement is completed, the device chip center coordinates are measured based on the device chip image information input from the first frame memory 156. When the reference measurement points managed by the reference measurement point and measurement area management unit 186 are processed one by one and the device chip center coordinates of all the reference measurement points are measured, the device chip position interpolation unit 184a performs the reference measurement. With reference to the measurement area data managed by the point and measurement area management unit 186, all device chip center coordinates on the WLCSP are calculated.

  The measurement region adaptation unit 188 determines whether or not the position matching degree between the calculated center coordinates of the device chip and the actually measured center coordinates falls within a required matching degree that is a matching degree required for measuring the electrical characteristics of the device chip. A determination unit 188a for determining the degree of coincidence and a measurement region subdivision unit 188b that further divides a measurement region including a reference measurement point that does not fall within the required degree of match are provided.

  The degree of coincidence of the device chip calculated in the image processing step and the center coordinates of the actually measured device chip are obtained by the degree of coincidence verification unit 188a, and the degree of coincidence required for measuring the electrical characteristics of the device chip is obtained. A verification step for determining whether or not a certain required degree of matching is satisfied is executed. The degree to which the required degree of matching is set depends on the size of the device chip and the like, so it is necessary to set it to the device type to be measured. The setting value of the required matching degree is also stored in the nonvolatile storage device 172, and the controller 170 can execute this program with reference to the nonvolatile storage device 172 when the verification step is executed. The

  Further, based on the test result in the test step, the measurement area subdivision unit 188b has a position match degree for a measurement area including a device chip whose position match degree is not within the required match degree range. Set four new reference points as reference measurement points and set the initial measurement area as four reference measurement points until three or more points do not line up in a straight line until the degree of match required to measure the electrical characteristics of the device chip A re-dividing step is performed in which the division is hierarchically divided into the divided areas surrounded by. After the verification step and the subdivision step are completed, they are written in the nonvolatile storage device 172 as a parameter set of product parameters.

  The frame reference point management unit 190 uses the position set in the dicing frame as a reference position, the reference measurement point coordinate management unit 190a for managing the device chip center coordinates measured as the reference measurement point based on the reference position, picking Thus, when a device chip does not exist at a position corresponding to the reference measurement point, another reference measurement point is selected as another reference measurement point and is managed.

  A device chip measured as a reference measurement point based on the reference position, with a position arbitrarily set on the dicing frame that fixes the dicing tape on which the device chip is placed fixed by the coordinate management unit 190a of the reference measurement point A coordinate management step for the reference measurement point for managing the center coordinates is executed. Further, another reference measurement point selecting unit 190b selects another reference measurement point as another reference measurement point when there is no device chip at the position corresponding to the reference measurement point, and adds another reference measurement point to the reference measurement point. A measurement point selection step is executed. After the coordinate management step and the separate reference measurement point selection step are completed, the newly set measurement area data and reference measurement point data are managed by the reference measurement point and measurement area management unit 186.

  The details of the above-described processing contents executed in the central processing unit 180 are as follows: 1. Alignment of singulated device chips and 2. Device chip array in a device chipped state after picker process Since it was explained in detail in the alignment, it will not be repeated. Further, the above-described processing executed in the central processing unit 180 can be executed by a normal personal computer.

  From the first arithmetic unit 160, device chip position coordinate information reflecting the above-described alignment result is supplied to the display unit 166 and the first data output unit 162. In addition, pattern matching processing result data is supplied from the image processing device 115 to the first arithmetic unit 160, and device chip testing result data from the device chip tester is supplied from the first data input unit 164. Further, the display device 166 gives an instruction necessary for executing the above-described processing.

  Referring to FIG. 41 again, the configuration of CSP prober control unit 130 will be described. The CSP prober control unit 130 includes a position information capturing unit 132 that captures the position coordinates of the measurement table 124, a second arithmetic unit 136, a second data output unit 138, a second data input unit 140, a third data output unit 144, 3 A data input unit 146 and a measurement control unit 142 for controlling the measurement table 124 are provided. The position information capturing unit 132 captures the position information sent from the measurement table 124, and the position information is sent to the second arithmetic unit 136.

  In addition to the position information of the measurement table 124, the second arithmetic unit 136 includes the position coordinates of the device chip from the first data output unit 162 provided in the device chip position controller 150 via the second data input unit 140. Data about is also sent. The second arithmetic unit 136 is a device that is input from the display unit 166 based on the position information of the measurement table 124, the position coordinates of the device chip, and the data related to the two-dimensional array position (position on the index map) of the device chip. According to the electrical characteristic measurement instruction of the chip, the position coordinate of the device chip to be measured is adjusted with respect to the position of the measurement electrode 122, the movement destination position of the measurement table 124 is determined, and the movement execution instruction is sent to the measurement control unit Send to 142. The device chip electrical characteristic measurement instruction is simultaneously sent as a separate route from the third data output unit 144 to the device chip tester 200 with the position on the index map of the device chip for measuring the electrical characteristics and the measurement execution instruction. The device chip electrical characteristic measurement instruction input from the display device 166, on the other hand, is controlled through the first arithmetic device 160, the first data output unit 162, the second data input unit 140, and the second arithmetic device 136. Part 142. On the other hand, the information is transmitted from the second arithmetic unit 136 to the device chip tester 200 via the third data output unit 144.

  In order to adjust the position coordinates of the device chip to be measured and the position of the measurement electrode 122, as described with reference to FIG. 32, FIG. 33 and FIG. The parallel movement and the rotational movement are calculated according to the number of chips by the least square method, the movement destination coordinates of the measurement table 124 are determined, and an instruction is issued from the measurement control unit 142 to the measurement table 124.

  The result of the electrical characteristic measurement is sent to the first arithmetic unit 160 via the device chip tester 200, the third data input unit 146, the second arithmetic unit 136, the second data output unit 138, and the first data input unit 164. . The result is converted by the first arithmetic unit 160 into a form that can be understood by the operator of the CSP prober who operates the display device 166, sent to the display device 166, and displayed.

  As shown in FIG. 30, the electrical characteristics of the device chip are measured while moving the index map in the order of step 1 and step 2. For example, in Fig. 30, 4 (2 x 2) multi-testing is performed, so testing is performed in units of one region (range enclosed by a broken line or solid square) in which two device chips are arranged in the vertical and horizontal directions. Will be carried out. That is, it is necessary to move the device chip with respect to the measurement electrode for each measurement unit. To this end, the measurement electrode is fixed, and the dicing frame on which the device chip is placed is moved by moving the measurement table.

  During device chip alignment, an instruction to move the measurement table 124 from the device chip position control device 150 to a position where the device chip to be measured can be captured by the camera 148 is sent to the measurement unit 120 via the CSP prober control unit 130. It is the composition to put out. The content of the instruction to move the measurement table 124 issued from the device chip position controller 150 is determined each time through an image processing step, a verification step, a subdivision step, a coordinate management step for a reference measurement point, a separate reference measurement point selection step, etc. Is done.

  On the other hand, at the time of measuring the electrical characteristics of the device chip, the CSP prober control unit 130 gives an instruction for the measurement electrode 122 to contact the device chip with the device electrode, to leave the device electrode, or to move the measurement position. In order to acquire the information for this from the device chip position control device 150, the following route is used.

  The instruction content about the movement of the measurement position determined in the first arithmetic device 160 is transmitted to the second arithmetic device 136 via the first data output unit 162 and the second data input unit 140. The second calculation device 136 calculates the movement destination position of the measurement table 124 in consideration of the position coordinates of the measurement table 124 at the present time, and the measurement position movement instruction content transmitted via the second data input unit 140 Is adapted and transmitted to the measurement control unit 142. Based on this movement information, the measurement control unit 142 sends a movement destination position of the measurement table 124 and a movement instruction signal to the measurement unit 120.

  In addition, when measuring the electrical characteristics of the device chip, information for the CSP prober control unit 130 to give an instruction for measuring the electrical characteristics to the device chip tester 200 is sent from the first arithmetic unit 160 to the first data Transmitted to the second arithmetic unit 136 via the output unit 162 and the second data input unit 140, and the second arithmetic unit 136 confirms the array position on the index of the device chip to be measured, and the second data The instruction about the measurement of the electrical characteristics transmitted through the input unit 140 is transmitted to the device chip tester 200 through the third data output unit 144.

  As described above, the CSP prober control unit 130 gives the measurement unit 120 an instruction to contact the device chip with the device electrode or to leave the device chip and to move the measurement position. In addition, the CSP prober control unit 130 has a function of acquiring information regarding these instruction contents from the device chip position control device 150. The device chip position control device 150 has a function of acquiring information such as the position of the device chip from the camera 148 and executing the device chip position measurement method of the present invention.

  Therefore, the device chip position measurement method of the present invention is realized by the device chip position measurement system built in the device chip position control device 150 with the assistance of the CSP prober control unit 130. That is, according to the CSP prober described with reference to FIG. 41 and FIG. 42, the individual center position coordinates of the device chip whose position is slightly displaced by being separated into pieces on the dicing tape and The inclination of the device chip can be accurately determined, and the electrical characteristics of the device chip can be measured after adjusting the relationship between the position of the device chip and the position of the measurement electrode of the tester based on the result.

  As described above, according to the device chip position measurement method of the present invention, the center coordinates of all the device chips that are not measured are calculated by measuring the center coordinates of the device chips set as the reference measurement points. Can do. The center coordinates of the device chip in the region surrounded by the selected reference measurement point by the image processing step can be obtained by calculation without directly measuring. Therefore, the measurement man-hours can be greatly reduced as compared with the method of measuring the total number of device chips and determining their positions.

  In the verification step, the calculated center coordinates of the device chip are compared with the measured center coordinates of the device chip, and it is determined whether or not the position match is within the match required for measuring the electrical characteristics of the device chip. Determined. Therefore, if the degree of match is within the required range, the process of measuring the electrical characteristics can proceed. That is, if the measurement result is a failure, it can be determined that it is a failure of the device chip, not due to a positional shift between the measurement electrode of the tester (installed in the CSP prober) and the device electrode of the device chip.

  On the other hand, if the degree of match is not within the required range, the next division step is executed. In the division step, the degree of coincidence between the calculated center coordinates and the measured center coordinates for the measurement region including the device chip that is not within the required degree of matching is measured for the electrical characteristics of the device chip. The measurement area is divided hierarchically into the area surrounded by the newly selected four points until it falls within the matching degree required for. As a result, the degree of matching between the measurement electrode of the device chip tester (installed in the CSP prober) and the device electrode of the device chip is within the required range, and finally the electrical characteristics of the device chip You can proceed to the measurement process.

  The device chip position measuring method of the present invention is suitable by using a normal CSP prober having a function capable of testing while a large number of device chips are simultaneously stuck on a dicing tape in a WLCSP process or the like.

It is a process figure with which it uses for description of singulation of a device chip. It is a figure with which it uses for description of the displacement of the device chip which generate | occur | produces when dividing a device chip into pieces. It is a figure with which it uses for description of the relationship between the expansion | extension amount of a dicing tape, and the amount of slack. It is a figure with which it uses for description of the arrangement | sequence of a device chip | tip when the dicing tape is expanded after separation. It is a figure which shows an example of the measurement electrode (pogo pin) of a tester, and a device electrode (pump terminal). It is a figure where it uses for description of the state which piled up and contacted the measurement electrode (pogo pin) of the tester on the device electrode (pump terminal). It is a figure with which it uses for description of the corresponding method in the case of testing the inclined device chip. It is a figure with which it uses for description of the position measurement of the device chip separated, and the interpolation method. It is a figure with which it uses for description of the position measurement of the device chip separated, and the interpolation method. It is a figure where it uses for description about the reference measurement position on WLCSP. It is a figure which shows an example of a WLCSP index map. It is a figure which uses for description of the effectiveness of the method of dividing | segmenting into the whole displacement and measurement area | region by individualization of the device chip on WLCSP. It is a figure where it uses for description of the process divided | segmented into a measurement area | region. It is a figure where it uses for description of the method of determining the inside and the exterior of an area | region. It is a figure where it uses for description of the relationship between a point and a quadrilateral. It is a figure where it uses for description of the method of adding a new area | region boundary point. It is a flowchart which shows the procedure of an initial measurement area | region division | segmentation step. It is a figure where it uses for description of the completion method of area | region formation and the determination method of an area | region boundary point. It is a flowchart which shows the procedure of a test | inspection process. It is a flowchart which shows the procedure of a subdivision step. It is a figure where it uses for description of the relationship between the area | region subdivided and the surrounding area | region. It is a flowchart of alignment of the separated WLCSP. It is a figure where it uses for description of the process for the angle adjustment of WLCSP. It is a figure where it uses for description of the extraction method of the position of the device chip which respond | corresponds with an index map. It is a figure where it uses for description of the encoding method of the presence or absence of an adjacent device chip. It is a figure where it uses for description of the setting method of an alternative reference | standard measurement point. It is a figure with which it uses for description of the coordinate management method on the basis of the feature point on a dicing frame. It is a figure with which it uses for description of the reference measurement point newly set by another reference measurement point selection step. It is a flowchart of the alignment of the tooth missing state WLCSP after picking. It is a figure where it uses for description of the process of testing. It is a figure where it uses for description of the center position of the measurement electrode of the tester of a device chip unit, and the center position of a device chip side terminal. It is a figure where it uses for description of the relationship between the center position of the measurement electrode of a tester, and the center position of a device chip side terminal. It is a figure which shows the state which the measurement electrode center position and the device chip center position overlapped best by performing parallel movement and rotational movement. It is a figure with which it uses for description of the overlap state of the measurement electrode center position and device chip center position in the case where rotation correction is performed and not. It is a figure where it uses for description of the influence of the inclination of the device chip side terminal with respect to the measurement electrode of a tester. It is a figure where it uses for description of the position of the measurement electrode in 2 multi-testing. It is a figure with which it uses for description of the position of the device chip side terminal in 2 multi-testing. It is a figure which shows the example of WLCSP with a small device chip size. It is a figure with which it uses for description of the measurement position for adjusting the angle of WLCSP on the measurement table in the case of a small device chip. It is a figure with which it uses for description of the reference | standard measurement point and measurement order in the case of a small device chip. It is a figure where it uses for description of the schematic structure of a CSP prober. FIG. 4 is a diagram for explaining a schematic configuration of a first arithmetic unit.

Explanation of symbols

10: WLCSP
30, 32, 40, 104: Device chip
12, 24, 50: Dicing frame
14, 26, 52: Dicing tape
20: Camera for alignment
38, 92, 106a, 106b: Tester measurement electrodes (pogo pins)
42, 94, 108a, 108b: Device electrode (pump)
54: Wafer
110: CSP prober controller
115: Image processing device
120: CSP prober measurement unit
122: Measuring electrode
124: Measurement table
130: CSP prober controller
132: Location information capture unit
136: Second arithmetic unit
138: Second data output section
140: Second data input section
142: Measurement control unit
144: Third data output section
146: Third data input section
148: Camera
150: Device chip position controller
152: Image capture unit
154: Image conversion processor
156: First frame memory
160: First arithmetic unit
162: First data output section
164: First data input section
166: Display section
170: Control unit
172: Nonvolatile storage device
180: Central processing unit
182: Initial measurement area division
184: Image processing unit
186: Reference measurement point and measurement area manager
188: Measurement area adaptation section
190: Frame reference point management department
200: Device chip tester

Claims (4)

In the two-dimensional array of device chips, as a point, the position on the array of any device chip is pointed out, and all the device chips are all measurement areas composed of the points, and three or more points are arranged on the same straight line. Set four or more points with no initial reference measurement points as polygons formed by connecting the four or more initial reference measurement points as the initial measurement region, and the initial measurement region where adjacent initial measurement regions touch each other An initial measurement area dividing step for dividing the entire measurement area;
The center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step are measured, and the center coordinates of the device chip and the device chips are two-dimensionally arranged on the matrix In this case, a projective transformation is defined to correspond to an index that is designated as a combination of two integers indicating the number of each row and column in which the device chip is located, and is formed with the four or more initial reference measurement points. An image processing step of calculating center coordinates of all device chips included in the initial measurement region using the projective transformation ;
A device chip position measuring method comprising: In a two-dimensional array of device chips, a point on the array of any device chip is indicated as a point, and all the device chips are all measurement areas composed of the points, and three or more points are aligned on the same line. Set four or more points with no initial reference measurement points as polygons formed by connecting the four or more initial reference measurement points as the initial measurement region, and the initial measurement region where adjacent initial measurement regions touch each other An initial measurement area dividing step for dividing the entire measurement area;
Measure the center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step, and include all the initial measurement areas included in the initial measurement area formed by the four or more initial reference measurement points. An image processing step for calculating the center coordinates of the device chip;
The center coordinates of the device chip calculated in the image processing step are compared with the measured center coordinates of the device chip, and the degree of coincidence of both positions is the degree of coincidence required for measuring the electrical characteristics of the device chip (“required” A test step for determining whether the degree of match is within a range of
Based on the test result in the test step, a new reference measurement point is added to the initial measurement area including the device chip that does not fall within the required degree of matching, and three or more points are aligned on a straight line. By dividing the initial measurement area into a plurality of measurement areas formed by four or more reference measurement points that do not have any of the four points or more until the matching degree of the position falls within the required matching degree. A device chip position measuring method comprising: a subdivision step for hierarchically dividing a divided area surrounded by reference measurement points including a reference measurement point. In a two-dimensional array of device chips, a point on the array of any device chip is indicated as a point, and all the device chips are all measurement areas composed of the points, and three or more points are aligned on the same line. Set four or more points with no initial reference measurement points as polygons formed by connecting the four or more initial reference measurement points as the initial measurement region, and the initial measurement region where adjacent initial measurement regions touch each other An initial measurement area dividing step for dividing the entire measurement area;
Measure the center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step, and include all the initial measurement areas included in the initial measurement area formed by the four or more initial reference measurement points. An image processing step for calculating the center coordinates of the device chip;
The center coordinates of the device chip calculated in the image processing step are compared with the measured center coordinates of the device chip, and the degree of coincidence of both positions is the degree of coincidence required for measuring the electrical characteristics of the device chip (“required” A test step for determining whether the degree of match is within a range of
Based on the test result in the test step, a new reference measurement point is added to the initial measurement area including the device chip that does not fall within the required degree of matching, and three or more points are aligned on a straight line. By dividing the initial measurement area into a plurality of measurement areas formed by four or more reference measurement points that do not have any of the four points or more until the matching degree of the position falls within the required matching degree. A subdivision step for hierarchically dividing the divided area surrounded by the reference measurement points including the reference measurement points that have been performed,
It has a function of storing data relating to the reference measurement point obtained in the four steps and data relating to the measurement region in a nonvolatile storage device, and is stored in the nonvolatile storage device instead of the initial measurement region dividing step. Loading data relating to a reference measurement point and said measurement area;
A device chip position measuring method comprising: In the two-dimensional array of device chips, as a point, the position on the array of any device chip is pointed out, and all the device chips are all measurement areas composed of the points, and three or more points are arranged on the same straight line. Set four or more points with no initial reference measurement points as polygons formed by connecting the four or more initial reference measurement points as the initial measurement region, and the initial measurement region where adjacent initial measurement regions touch each other An initial measurement area dividing step for dividing the entire measurement area;
The center coordinates of the device chip located at all the initial reference measurement points set in the initial measurement area dividing step are measured, and the center coordinates of the device chip and the device chips are two-dimensionally arranged on the matrix In this case, a projective transformation is defined to correspond to an index that is designated as a combination of two integers indicating the number of each row and column in which the device chip is located, and is formed with the four or more initial reference measurement points. An image processing step of calculating center coordinates of all device chips included in the initial measurement region using the projective transformation ;
A position arbitrarily set on a dicing frame to which the dicing tape on which the device chip is placed is fixed is set as a reference position, and a center of a specific device chip is selected from the device chips on the dicing tape based on the reference position. A reference measurement point coordinate management step for managing reference measurement point coordinates for specifying a position;
When there is no device chip at a position corresponding to the reference measurement point, and when all the device chips in the region cannot be covered with a polygon formed by connecting the reference measurement points, as another reference measurement point Selecting another reference measurement point and adding another reference measurement point
A device chip position measuring method comprising:

Images