Wafer bonding strength testing method

Abstract

The invention discloses a wafer bonding strength testing method, which comprises the following steps of: cutting a bonded wafer into strip-shaped samples, adjusting a precise translation table to enable a laser head to be aligned with a bonding layer, scanning the bonding layer of the sample by using laser to obtain an inward initial crack, monitoring the position of the initial crack by using a microscope, and adjusting the precise translation table according to the position of the initial crack to test the bonding strength of the wafer. The method comprises the following steps: calibrating a crack to obtain a calibrated crack, adhering test pull rings to two sides of a sample with the calibrated crack, performing tension test on the pull rings to obtain a load-displacement curve of the sample, and calculating the bonding strength of the sample according to the load-displacement curve of the sample. According to the method, the submicron aligned preset cracks are processed through cutting and closed-loop calibration, and then the wafer bonding strength is quantitatively tested through a DCB method, so that the problem that the bonding strength of internal positions of wafers of a D2W bonding sample and a W2W bonding sample cannot be quantitatively tested is solved.

Description

Technical Field

[0001] The invention relates to a wafer bonding strength testing method and belongs to the technical field of integrated circuit manufacturing. Background Art

[0002] Hybrid bonding (HB) is a technology that achieves vertical interconnection between Cu and dielectric materials at low temperatures through plasma activation (solid-state bonding). It offers advantages such as small interconnect pitch, minimal thermal impact, and low-resistance interconnection. During HB process development, precise quantitative measurement of bond strength is essential to evaluate process performance, including copper plating, chemical mechanical polishing, cleaning, plasma activation, and annealing, as well as the service reliability of hybrid bonded samples.

[0003] Currently, the main bond strength testing methods include the double cantilever beam (DCB)-based blade insertion test and the Czochralski method. The blade insertion method is only suitable for wafer-to-wafer (W2W) bonding with a bevel gap, and is not suitable for die-to-wafer (D2W) or die-to-die (D2D) bonding. Furthermore, the blade insertion method cannot characterize the center bond strength of W2W bonding. The Czochralski method can easily cause bulk silicon fracture when testing samples with high bond strength, making it impossible to accurately characterize bond strength.

[0004] Another DCB testing method is to introduce a pre-crack through gold plating or laser scanning before bonding. After bonding, tension is applied to the upper and lower beams on either side of the pre-crack to induce delamination. However, gold plating is time-consuming and costly, while laser-induced cracking increases roughness, thus affecting bonding. Both methods are unsuitable for testing finished samples. Summary of the Invention

[0005] Purpose: Chip-to-wafer (D2W) bonding products are increasingly used and play a crucial role in high-computing chip applications. Because the bond interface of D2W products lacks the aforementioned chamfered structure, conventional blade insertion methods cannot be used to measure bond strength. To overcome these shortcomings in the prior art, the present invention provides a wafer bond strength testing method.

[0006] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A wafer bonding strength testing method comprises the following steps:

[0008] Step 1: Cut the bonded wafer pair into strip samples, and polish the strip samples to obtain polished samples.

[0009] Step 2: Fix the polished sample on the fixture above the precision translation stage and adjust the precision translation stage so that the laser head is aligned with the bonding layer.

[0010] Step 3: Use laser to scan the bonding layer of the sample to obtain the initial inward crack.

[0011] Step 4: Monitor the initial crack position through a microscope and record the distance between the crack tip and the bonding layer on the left and right sides of the sample. If the distance between the crack tip and the bonding layer on the left and right sides of the sample is within the alignment accuracy threshold, proceed to step 6. If one of the distances between the crack tip and the bonding layer on the left and right sides of the sample is greater than the alignment accuracy threshold, calculate the angle of deviation between the bonding layer and the laser processing trajectory based on the distance between the crack tip and the bonding layer on the left and right sides of the sample, and adjust the position of the sample on the precision translation stage according to the deviation angle.

[0012] Step 5: Remove the misaligned cracks by grinding and polishing, and repeat steps 3 and 4 until the distances between the crack tips and the bonding layer on the left and right sides of the sample are within the alignment accuracy threshold.

[0013] Step 6: Glue test rings on both sides of the sample, perform a tensile test on the rings, obtain the load-displacement curve of the sample, and calculate the bonding strength of the sample based on the load-displacement curve of the sample.

[0014] Optionally, the method of cutting the bonded wafer pair is one of mechanical cutting, laser stealth cutting, plasma cutting, chemical cutting and laser induced plasma cutting.

[0015] Optionally, the polishing method includes one of mechanical polishing, chemical mechanical polishing and plasma polishing.

[0016] Optionally, the polishing liquid type includes one or more of silicon dioxide, aluminum oxide, cerium oxide, and diamond.

[0017] Optionally, the sandpaper type includes one or more of aluminum oxide sandpaper, silicon carbide sandpaper, diamond sandpaper, zirconium corundum sandpaper, and ceramic sandpaper.

[0018] Optionally, the precision translation stage includes one of a manual translation stage and an automatic translation stage.

[0019] Optionally, the deviation angle can be expressed as follows:

[0020]

[0021] in, is the angle of deviation, is the offset distance between the crack tip and the bonding layer on one side of the sample, is the offset distance between the crack tip and the bonding layer on the other side of the sample, is the width of the cut surface of the sample.

[0022] Optionally, step 6 includes:

[0023] Under the atmosphere, the test rings are driven to move by a precision tensile testing machine.

[0024] If a single DCB test is performed, the precision tensile testing machine will continue to move without resetting until the bonding layer of the sample is separated. The load of the precision tensile testing machine and the displacement of the bonding layer separation will be recorded to obtain a load-displacement curve.

[0025] Based on the load-displacement curve, calculate the slope of the load-displacement curve.

[0026] The modulus is calculated from the slope of the load-displacement curve.

[0027] Calculate bond strength based on modulus and peak load.

[0028] Optionally, step 6 includes:

[0029] Step 6.1: Under atmosphere, use a precision tensile testing machine to drive the test rings to move.

[0030] Step 6.2: If a cyclic DCB test is performed, when the load of the precision tensile testing machine reaches its peak and continues to move for a certain distance, the tensile testing machine is reset and the load of the precision tensile testing machine and the displacement of the bonding layer are recorded to obtain the load-displacement curve.

[0031] Step 6.3: Repeat step 6.2 until the sample is completely separated and several load-displacement curves are obtained.

[0032] Step 6.4: Based on each load-displacement curve, calculate the slope of each load-displacement curve.

[0033] Step 6.5: Calculate the modulus corresponding to each slope based on the slope of each load-displacement curve.

[0034] Step 6.6: Calculate the bond strength for each cycle based on the modulus and peak load corresponding to each slope.

[0035] Step 6.7: Calculate the average value of the bonding strength in each cycle as the bonding strength.

[0036] Optionally, the precision tensile testing machine is a constant speed displacement machine with a speed of 0.001-100 mm / min.

[0037] Beneficial Effects: The wafer bond strength testing method provided by this invention solves the technical problem that conventional blade insertion methods rely on the special chamfered structure of the wafer edge, can only test the bond strength of wafer-to-wafer (W2W) bonded products and only the edge area of ​​such products. Furthermore, it cannot measure the bond strength of the inner area of ​​the wafer.

[0038] The present invention provides a wafer bond strength testing method that uses a cutting module and a closed-loop calibration unit (including a microscope and a translation stage) to process submicron-level aligned pre-set cracks, and then quantitatively tests the wafer bond strength using the DCB method. This solves the problem of being unable to quantitatively test the bond strength inside the wafer of D2W bonded samples and W2W bonded samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The figure is a flow chart of a wafer bonding strength testing method of the present invention.

[0040] Figure 2 Schematic diagram of wafer pair cutting.

[0041] Figure 3 Schematic diagram of the relationship between laser scanning cracks, cracks and bonding layer positions. Figure 3 (a) is a schematic diagram of the laser scanning the bonding layer for the first time. Figure 3 (b) is a schematic diagram of a laser crack where the crack tip does not overlap with the bonding layer. Figure 3 (c) is a schematic diagram of the distance between the crack tip and the bonding layer on both sides of the sample observed by microscope. Figure 3 (d) is a schematic diagram of grinding to remove misaligned cracks. Figure 3 (e) is a schematic diagram of the laser scanning the bonding layer again. Figure 3 (f) is a schematic diagram of a laser crack with deviation in the horizontal direction. Figure 3 (g) is a schematic diagram of the distance between the crack tip and the bonding layer observed by microscope. Figure 3 (h) is a schematic diagram of grinding to remove misaligned cracks. Figure 3 (j) is a schematic diagram of the laser scanning the bonding layer again. Figure 3 (k) is a schematic diagram showing that the crack tip and the bonding layer are fully aligned.

[0042] Figure 4 This is a magnified schematic diagram of the crack tips on both sides of the sample observed under a microscope, where: Figure 4 (a) is a schematic diagram of the crack tip and bonding layer on the left. Figure 4 (b) is a schematic diagram of the crack tip and bonding layer on the right.

[0043] Figure 5 Schematic diagram of the DCB test sample.

[0044] Figure 6 This is an enlarged schematic diagram of the crack tips on both sides of Example 2, where: Figure 6 (a) is a schematic diagram of the crack tip and bonding layer on the left. Figure 6 (b) is a schematic diagram of the crack tip and bonding layer on the right.

[0045] Figure 7 This is a schematic diagram of the DCB test process, where: Figure 7 (a) is a schematic diagram of the sample being subjected to tension on both sides. Figure 7 (b) is a schematic diagram of the cantilever beam starting to bend under the action of tension. Figure 7 (c) is a schematic diagram showing that the elastic potential energy accumulated by the bending of the cantilever beam exceeds the bonding energy, and the bonding interface is opened. Figure 7 (d) is a schematic diagram showing that the bonding interface is opened and then moves for a distance before resetting. At this time, the cantilever beam does not bend. Figure 7 (e) is a schematic diagram showing that the sample is subjected to tension again and the cantilever beam bends again.

[0046] Figure 8 is the DCB load-displacement curve, where Figure 8 (a) Schematic diagram of the load-displacement curve of a single DCB test. Figure 8 (b) Schematic diagram of the load-displacement curve of the cyclic DCB test. DETAILED DESCRIPTION

[0047] The following is a clear and complete description of the technical solutions in the examples of the present invention, in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.

[0048] The present invention will be further described below with reference to specific embodiments.

[0049] Example 1:

[0050] This embodiment introduces a wafer bonding strength testing method. Figure 1 As shown, the following steps are included:

[0051] Step 1: Cut the bonded wafer pair into strip samples to facilitate testing the bond strength at a target location on the wafer or at different locations on the wafer, thereby characterizing the overall bonding uniformity. The strip samples are then polished to ensure that the sides are flat and smooth, and the bond layer is clearly visible, resulting in a polished sample. Polishing methods include mechanical polishing, chemical mechanical polishing, and plasma polishing.

[0052] Step 2: Fix the polished sample on the fixture above the precision translation stage, and change the position of the sample by adjusting the precision translation stage so that the laser head is aligned with the bonding layer.

[0053] Step three: Use laser to scan the bonding layer of the sample to obtain the initial inward crack.

[0054] In step 4, the initial crack position is monitored using a microscope, and the distances between the crack tips and the bonding layer on the left and right sides of the sample are recorded. If both distances are within the alignment accuracy threshold, the process proceeds to step 6. If either distance is greater than the alignment accuracy threshold, the angle of deviation between the bonding layer and the laser machining trajectory is calculated based on the distances between the crack tips and the bonding layer. The position of the sample on the precision translation stage is adjusted based on the deviation angle. Microscopes include optical microscopes, scanning electron microscopes, and transmission electron microscopes.

[0055] Step 5: Remove the misaligned cracks by grinding and polishing, and repeat steps 3 and 4 until the distances between the crack tips and the bonding layer on the left and right sides are within the alignment accuracy threshold.

[0056] Step six, stick test pull rings on both sides of the sample with standard test cracks. In an atmosphere of air, high-purity nitrogen, high-purity argon, etc., perform a tensile test on the sample with the pull ring using a precision tensile testing machine. The tensile testing mode is constant speed displacement, and the displacement speed is 0.001-100 mm / min. Whenever the load in the test curve reaches a peak and continues to displace a certain distance, the tensile testing machine is reset to obtain a load-displacement curve, and the tensile test is repeated until the sample is completely separated. At this time, several load-displacement curves are obtained, and the curve data is processed, and the modulus is obtained according to the slope of the curve. The bond strength is calculated based on the modulus and peak load, and the average bond strength of the last several load-displacement curves is taken to characterize the bond strength of the sample.

[0057] Furthermore, in step one, it includes:

[0058] Step 101: Cut the bonded wafer pair into long strip samples (hereinafter referred to as samples), the length and width of the samples are less than 300 mm, and the thickness is less than 3 mm. Figure 2 As shown, a bonded wafer pair consisting of two wafers 1 is cut by a cutting blade 4 .

[0059] Step 102: polish the cut surface of the sample.

[0060] Furthermore, in step 101, the method of cutting the bonded wafer pair is one of mechanical cutting, laser stealth cutting, plasma cutting, chemical cutting, and laser induced plasma cutting.

[0061] Furthermore, in step 102, the polishing method includes mechanical polishing, chemical mechanical polishing and plasma polishing.

[0062] Furthermore, in step 102, the polishing liquid used includes one or more materials such as silicon dioxide, aluminum oxide, cerium oxide, and diamond.

[0063] Furthermore, in step 102 , the sandpaper used includes one or more of aluminum oxide sandpaper, silicon carbide sandpaper, diamond sandpaper, zirconium corundum sandpaper, and ceramic sandpaper.

[0064] Furthermore, in step 2, the precision translation stage used is one of a manual translation stage and an automatic translation stage.

[0065] Furthermore, in step three, if Figure 3 As shown in (a), including:

[0066] Step 301 : Using laser cutting, plasma cutting, chemical cutting or a cutting laser 7 of an invisible laser cutting machine to machine a laser-induced crack 3 in the bonding layer 2 .

[0067] Furthermore, in step 4, it includes:

[0068] Step 401: Use one of an optical microscope, an electron microscope, and an X-ray microscope to measure the offset distance between the crack tip and the bonding layer on both sides of the sample, and calculate the sample offset angle, such as Figure 3 As shown in (b), the calculation formula of the sample offset angle is as follows:

[0069]

[0070] in, is the sample offset angle, is the offset distance between the crack tip and the bonding layer on one side of the sample, is the offset distance between the crack tip and the bonding layer on the other side of the sample, is the width of the cut surface of the sample.

[0071] Step 402: According to the sample offset angle, adjust the angle of the precision translation stage, such as Figure 3 As shown in (c).

[0072] Furthermore, in step five, the method includes:

[0073] Step 501: Remove the misaligned cracks, re-process the cracks, repeat the offset distance measured in step 4, and adjust the horizontal position of the precision translation stage, such as Figure 3 As shown in (d), (e) and (f).

[0074] Step 502: Repeat step 3, as Figure 3 As shown in (g), (j), and (k), the final sample with a standard test crack 9 is obtained. The specific structure is as follows Figure 4 shown.

[0075] Furthermore, in step six, it includes:

[0076] Step 601: Adhere the test ring 10 to the periphery of the crack of the sample. Figure 5 shown.

[0077] Step 602: Test the sample with the pull ring from step 601 using a precision tensile testing machine in an atmosphere such as air, high-purity nitrogen, or high-purity argon. The tensile testing machine is set to constant displacement mode at a displacement rate of 0.001-100 mm / min. Each time the load in the test curve reaches a peak, the sample is displaced for a certain distance and then reset to obtain a load-displacement curve.

[0078] Step 603: Repeat step 602 until the sample is completely separated. At this point, several load-displacement curves are obtained. The load-displacement curve data is processed, and the slope of each curve is fitted to obtain the modulus. The bond strength is calculated based on the modulus and peak load.

[0079] Example 2:

[0080] This embodiment provides a method for testing wafer bonding strength, including the following steps:

[0081] Step 1: Cut the bonded wafer pair into long strip samples and polish them;

[0082] In step one, include:

[0083] 11) Cut the bonded wafer pair into 5x40 mm samples.

[0084] The method of cutting the bonded wafer pair is one of mechanical cutting, laser stealth cutting, plasma cutting, chemical cutting, and laser induced plasma cutting.

[0085] 12) Polish the cut surface of the sample, with the last polishing direction perpendicular to the bonding layer.

[0086] Polishing methods include mechanical polishing, chemical mechanical polishing and plasma polishing;

[0087] The polishing liquid used includes one or more of silica-based, alumina-based, ceria-based, and diamond-based;

[0088] The types of sandpaper used include one or more of aluminum oxide sandpaper, silicon carbide sandpaper, diamond sandpaper, zirconium corundum sandpaper, and ceramic sandpaper.

[0089] Step 2: Monitor the crack position through a microscope, adjust the sample position with a precision translation stage, and perform initial positioning of the bonding interface of the sample to be processed.

[0090] The precision translation stage used is one of a manual translation stage and an automatic translation stage.

[0091] In this embodiment, the precision translation stage used is a manual translation stage, with a translation step accuracy of better than 1 mm in the XYZ direction and a rotation accuracy of better than 5' in the R direction.

[0092] Step three: Process an inward initial crack at the bonding interface. According to the offset of the initial crack and the bonding layer, adjust the precision translation stage to achieve high-precision alignment between the crack processing position and the bonding interface, where the alignment accuracy is better than 1 mm.

[0093] In step four, include:

[0094] 41) Use laser cutting, plasma cutting, chemical cutting or invisible laser cutting machine to process initial cracks at the bonding layer to accurately initiate cracking of the bonding interface.

[0095] In this embodiment, the laser cutting machine is a YAG laser cutting machine, and the cutting parameters are a speed of 10 mm / s, a current of 2 A, a frequency of 20 KHz, and a pulse width of 40 μs;

[0096] 42) Use one of the optical microscope, electron microscope and X-ray microscope to measure the offset distance between the crack tip and the bonding layer on both sides of the sample in step 2, and calculate the sample offset angle, such as Figure 6 As shown in (a) and (b), the offset distances between the crack tip and the bonding layer on both sides are 10μm and 6μm, respectively. Combined with the fact that the width of the sample is about 4900μm, the sample offset angle is about 0.187°.

[0097] 43) Adjust the angle of the precision translation stage according to the sample offset angle.

[0098] 44) The misaligned cracks were removed by sandpapering and the cracks were machined again using the laser.

[0099] 45) Adjust the horizontal position of the precision translation stage based on the offset distance measured in step 44).

[0100] Step 5: Sandpaper is used to remove the misaligned cracks, and the bonding layer of the sample is scanned twice by laser to produce a standard test crack.

[0101] Step 6: Perform a double cantilever beam test based on the calibrated crack and calculate the bond strength G based on the load-displacement curve. c .

[0102] In step six, include:

[0103] 61) Glue a test ring around the crack in the sample.

[0104] 62) Test the sample with the pull ring from step 61) using a precision tensile testing machine in an atmosphere of air, high-purity nitrogen, or high-purity argon. The tensile testing machine is in constant displacement mode with a displacement rate of 0.001-100 mm / min.

[0105] If a single DCB test is performed, that is, the push-pull test machine keeps moving without resetting, a cyclic DCB test is performed to accurately calculate the bond strength. Figure 7 As shown in (a) to (e), every time the load in the test curve reaches the peak and continues to move a certain distance, the tensile machine is reset to obtain a load-displacement curve, such as Figure 8 (a) shows the load-displacement curve. This curve can be divided into three segments: a, b, and c. In segment a, the cantilever beam bends, accumulating elastic potential energy, and the load increases linearly. In segment b, the energy accumulated by the bending of the cantilever beam exceeds the interfacial bonding energy, causing crack growth and a decrease in load. In segment c, the crack growth rate and the force gauge test speed reach a relative equilibrium. The curve in segment c, where the crack is steadily growing, is typically used to calculate bond strength.

[0106] Repeat the above process until the sample is completely separated, and then several load-displacement curves are obtained, such as Figure 8 (b)

[0107] The curve data is processed and the slope of the curve is used to obtain the modulus. The bond strength is calculated based on the modulus and peak load. The average bond strength of the last load-displacement curve is taken to represent the bond strength of the sample.

[0108] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A wafer bonding strength testing method, characterized by: The following steps are involved: Step 1: Cut the bonded wafer pair into strip samples, and polish the strip samples to obtain polished samples; Step 2: Fix the polished sample on the fixture above the precision translation stage and adjust the precision translation stage so that the laser head is aligned with the bonding layer; Step 3: Use laser to scan the bonding layer of the sample to obtain the initial inward crack; Step 4: Monitor the initial crack position through a microscope and record the distance between the crack tip and the bonding layer on the left and right sides of the sample. If the distance between the crack tip and the bonding layer on the left and right sides of the sample is within the alignment accuracy threshold, proceed to step 6. If one of the distances between the crack tip and the bonding layer on the left and right sides of the sample is greater than the alignment accuracy threshold, calculate the angle of deviation between the bonding layer and the laser processing trajectory based on the distance between the crack tip and the bonding layer on the left and right sides of the sample, and adjust the position of the sample on the precision translation stage based on the deviation angle. Step 5: Remove the misaligned cracks by grinding and polishing, and repeat steps 3 and 4 until the distances between the crack tips and the bonding layer on the left and right sides of the sample are within the alignment accuracy threshold; Step 6: Glue test pull rings on both sides of the sample, perform a tensile test on the pull rings, obtain the load-displacement curve of the sample, and calculate the bonding strength of the sample based on the load-displacement curve of the sample.

2. The wafer bonding strength testing method according to claim 1, wherein: The method of cutting the bonded wafer pair is one of mechanical cutting, laser stealth cutting, plasma cutting, chemical cutting and laser induced plasma cutting.

3. The wafer bonding strength testing method according to claim 1, wherein: The polishing method includes one of mechanical polishing, chemical mechanical polishing and plasma polishing.

4. The wafer bonding strength testing method according to claim 1, wherein: The polishing liquid type includes one or more of silicon dioxide, aluminum oxide, cerium oxide, and diamond.

5. The wafer bonding strength testing method according to claim 1, wherein: The sandpaper types include one or more of aluminum oxide sandpaper, silicon carbide sandpaper, diamond sandpaper, zirconium corundum sandpaper, and ceramic sandpaper.

6. The wafer bonding strength testing method according to claim 1, wherein: Precision translation stages include manual translation stages and automatic translation stages.

7. The wafer bonding strength testing method according to claim 1, wherein: The expression for the angle of deviation is as follows: ； in, is the angle of deviation, is the offset distance between the crack tip and the bonding layer on one side of the sample, is the offset distance between the crack tip and the bonding layer on the other side of the sample, is the width of the cut surface of the sample.

8. The wafer bonding strength testing method according to claim 1, wherein: The step 6 comprises: Under the atmosphere, the test rings are driven to move by the precision tensile testing machine; If a single DCB test is performed, the precision tensile testing machine will continue to move without resetting until the bonding layer of the sample is separated. The load of the precision tensile testing machine and the displacement of the bonding layer separation will be recorded to obtain the load-displacement curve. According to the load-displacement curve, calculate the slope of the load-displacement curve; The modulus was calculated from the slope of the load-displacement curve; Calculate bond strength based on modulus and peak load.

9. The wafer bonding strength testing method according to claim 1, wherein: The step 6 comprises: Step 6.1: Under atmosphere, use a precision tensile testing machine to drive the test rings to move; Step 6.2: If a cyclic DCB test is performed, when the load of the precision tensile testing machine reaches its peak, continue to move for a certain distance, reset the tensile testing machine, and record the load of the precision tensile testing machine and the displacement of the bonding layer to obtain the load-displacement curve; Step 6.3: Repeat step 6.2 until the sample is completely separated and several load-displacement curves are obtained; Step 6.4: Based on each load-displacement curve, calculate the slope of each load-displacement curve; Step 6.5: Calculate the modulus corresponding to each slope based on the slope of each load-displacement curve; Step 6.6: Calculate the bond strength for each cycle based on the modulus and peak load corresponding to each slope. Step 6.7: Calculate the average value of the bonding strength in each cycle as the bonding strength.

10. A wafer bonding strength testing method according to claim 8 or 9, characterized in that: The precision tensile testing machine has a constant displacement speed of 0.001-100 mm / min.

Images