Solid chip inspection method

Abstract

To distinguish a normal device from an abnormal device so that the device, which may possibly have defects such as chips and cracks because it is so small and brittle that visual inspection is not easy, receives little damage during inspection.SOLUTION: A solid chip inspection method, in which defects of a solid chip are detected and it is determined whether the chip is normal or abnormal, comprises: a step in which the solid chip is adhered to a vibration material through an adhesion layer; a step in which a vibration of the solid chip is detected while vibrating the vibration material; a step in which the vibration is converted to a frequency signal; a main component analysis step in which a first main component and a second main component of the solid chip are calculated by the main component analysis of the frequency signal; and a determination step in which the solid chip can be determined to be normal or abnormal by comparing a main component plot, which shows an area where the normal chip exists and an area where the abnormal chip exists in the first and second main components, with the first and second main components of the solid chip.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a method for inspecting solid-state chip devices, and more particularly to an inspection method for detecting damage to a solid-state chip before it is mounted on a circuit board or the like. [Background technology]

[0002] Many proposals have been made for inspection equipment for small chip-like devices such as semiconductors (hereafter referred to as "solid-state chips"). Equipment that performs electrical tests, which are usually conducted before mounting on a circuit board, is known. On the other hand, for defects or scratches on solid-state chips, visual inspection is relied upon using three-sided visual inspection equipment.

[0003] Defects and scratches have been detected by vibrating the object to be inspected and analyzing the vibration frequency. Patent Document 1 discloses a method in which an electromagnetic ultrasonic wave is irradiated onto the object to be inspected, laser ultrasonic waves are applied to the location, and defects, plate thickness, etc. of the object to be inspected are detected based on the reflected signal. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 177447 / 1983 Summary of the Invention [Problem to be solved by the invention]

[0005] The method of Patent Document 1 is considered to be effective for relatively large objects such as gas pipes, etc. However, since ultrasonic waves are used as a vibration means, it cannot be said to be suitable as a vibration means for small solid chips of 1 mm square or less, because the irradiation range becomes wide.

[0006] One possible method of excitation is to use a laser, but excitation using a laser requires high laser output, and there is a problem that the laser itself can damage the solid chip. [Means for solving the problem]

[0007] The present invention has been conceived in view of the above problems, and provides a method for detecting defects or scratches in a solid-state chip itself, without directly vibrating the solid-state chip being the test subject with a laser.

[0008] More specifically, the solid chip inspection method according to the present invention comprises the steps of: A method for inspecting a solid chip for detecting defects in the solid chip and judging whether the solid chip is normal or defective, comprising the steps of: Adhering the solid chip to a vibration material via an adhesive layer; detecting vibration of the solid-state tip while vibrating the vibration member; converting the vibration into a frequency signal; The method further comprises a step of determining whether the solid-state chip is normal or abnormal based on the frequency signal. Effect of the Invention

[0009] The solid-state chip inspection method of the present invention has the advantage of not damaging the solid-state chip itself, since the solid-state chip is placed on a vibrating material, the vibrating material is excited to cause vibration to the solid-state chip, and the vibration is analyzed to determine whether there is an abnormality in the solid-state chip.

[0010] Furthermore, if the process of determining whether the solid-state chip is normal or abnormal based on the frequency signal is performed by a principal component analysis process of performing principal component analysis on the frequency signal to calculate the first principal component and the second principal component of the solid-state chip, and determining whether the solid-state chip is normal or abnormal by comparing a principal component plot showing the range in the first principal component and the second principal component where normal products exist and the range in the first principal component and the second principal component where abnormal products exist with the first principal component and the second principal component of the solid-state chip, there is an effect that by storing the principal component plot in advance, it can be quickly determined whether the object to be inspected is normal or abnormal. [Brief description of the drawings]

[0011] [Figure 1] 1 is a diagram showing a configuration of a solid-state chip testing device for performing a solid-state chip testing method according to the present invention; [Diagram 2] FIG. 13 is a diagram showing another configuration of the solid-state chip testing device. [Diagram 3] FIG. 1 is a flow chart showing a process flow of a solid-state chip inspection method. [Figure 4] 1 is an example of a principal component plot. [Diagram 5] FIG. 11 is a flowchart showing the process of creating a principal component plot. [Figure 6] FIG. 2 is a diagram showing the internal configuration of a controller. [Figure 7] FIG. 13 is a diagram showing the vibration measurement situation in the embodiment and the results of Fourier transform of a normal product and an abnormal product. [Figure 8] FIG. 13 is a diagram showing the vibration measurement situation in the embodiment and the results of Fourier transform of a normal product and an abnormal product. [Figure 9] This is an example of creating a principal component plot. [Figure 10] This is an example of creating a principal component plot. [Figure 11] FIG. 13 is a diagram showing the difference in principal component plots between a case where a solid tip is directly vibrated and a case where the method of the present invention is used. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] The solid-state chip inspection method according to the present invention will be described below with reference to the drawings and examples. Note that the following description does not limit the present invention. The following description can be modified without departing from the spirit and scope of the present invention.

[0013] <Hardware configuration> 1 shows the configuration of a solid chip inspection device 1 for implementing a solid chip inspection method according to the present invention. The solid chip inspection device 1 according to the present invention comprises a vibrator 10, a vibration measuring device 12, and a controller 14. The object to be inspected is designated by the reference numeral 100. The object to be inspected 100 is bonded to a vibration material 104 via an adhesive layer 102.

[0014] The vibrator 10 vibrates not the object under test 100 itself, but the diaphragm 104 to which the object under test 100 is fixed. The vibrator 10 is not particularly limited as long as it can vibrate the diaphragm 104, but an excitation laser LL is preferable. For example, an element that vibrates by itself, such as a piezoelectric element, has a resonant frequency depending on the size and shape of the element, and may be subject to unexpected restrictions on the frequency used, etc. However, this does not exclude self-oscillating bodies such as piezoelectric elements.

[0015] Moreover, since the vibrator 10 does not directly vibrate the test object 100 (solid chip), it may be a high-output laser. If it is a laser, the frequency of the vibration can be easily controlled.

[0016] The vibration measuring device 12 measures the vibration of the test object 100 while the vibrator 10 vibrates the vibration material 104. A laser Doppler vibrometer is preferably used for this. Since it is non-contact, it does not interfere with the vibration of the test object 100. Furthermore, since the measurement spot can be narrowed, even a small solid chip can be irradiated. Furthermore, since the laser output is weak, it does not damage the test object 100. However, a microphone, acceleration sensor, etc. may also be used as long as the effect of the present invention can be obtained.

[0017] The controller 14 is composed of a microprocessor unit, a memory, an input / output device, etc. The controller 14 controls the vibration of the vibrator 10 by command signal Cv. In this control, the vibrating material 104 is vibrated at a plurality of frequencies. The controller 14 also operates the vibration measuring device by command signal Cm, and obtains vibration measurement values ​​by notification signal Sm. The obtained measurement values ​​are Fourier transformed and principal component analyzed by the controller 14, as will be described later.

[0018] FIG. 2 shows another configuration of a solid chip inspection device (hereinafter referred to as "solid chip inspection device 2"). The difference between solid chip inspection device 1 and solid chip inspection device 2 is that a vibration auxiliary material 16 is placed under the vibration material 104. The vibration material 104 and the vibration auxiliary material 16 do not need to be bonded together. Furthermore, in solid chip inspection device 2, the vibrator 10 vibrates the vibration auxiliary material 16. As will be shown in the examples described later, the vibration auxiliary material 16 can increase the vibration frequency of the test object 100. Subsequent signal processing is easier if the vibration frequency is on the order of several hundred to several kHz.

[0019] In this way, in the present invention, either the diaphragm 104 or the vibration support material 16 in a state where the diaphragm 104 and the vibration support material 16 are laminated is vibrated. Therefore, the state where the diaphragm 104 and the vibration support material 16 are laminated may be called the diaphragm 104. This may be said to be that the diaphragm 104 in contact with the test object 100 via the adhesive layer 102 is vibrated.

[0020] The solid chips of the object to be inspected 100 are mainly those in which a conductive layer is disposed on a plate-shaped ceramic such as silicon or zirconia. Solid chips include semiconductor chips and compound semiconductors such as indium phosphide. These are hard but brittle, and chips tend to occur on the cut surface. They may also crack. Such cracks include cleavage cracks resulting from the crystal structure of the material. The present invention identifies shape defects such as chips and cracks in solid chips as abnormal products.

[0021] It is preferable that the adhesive layer 102 is a layer that maintains its viscosity rather than one that hardens and solidifies. This is because the solid chip (test subject 100) will eventually be peeled off from the adhesive layer 102 and used for its intended purpose. In that case, it is preferable that it peels off easily. The adhesive layer 102 also absorbs vibrations from the vibrating material 104 and the vibration assist material 16 and transmits them to the solid chip. This prevents defects in the solid chip caused by vibrations during testing.

[0022] A flexible base film or the like is preferably used as the vibration material 104. This can also be said to be a mount for transporting a solid chip during the manufacture of the solid chip. For example, polymeric materials such as PET, PVC, and PO can be preferably used as the flexible film.

[0023] The vibration support material 16 is made of a material having a higher hardness than the vibration material 104. Plastic, metal plate, or the like can be suitably used. Furthermore, the device of the present invention can be realized regardless of the device's external appearance as long as it is equipped with the above features. In other words, it is not limited to being used as an individual inspection device, but may be mounted on an electrical inspection device, an external inspection device, a robot used in an assembly process, or the like and utilized.

[0024] <Processing flow> Next, the flow of the process in the present invention will be described. Please refer to FIG. 3. First, the test object 100 (solid chip) is prepared to be fixed to the oscillating material via an adhesive layer. At this time, the oscillating material does not have to be a part of the solid chip testing device 1 (or 2). In other words, a solid chip fixed to the oscillating material may be supplied as the test object 100. This corresponds to a process of adhering the solid chip to the oscillating material via an adhesive layer.

[0025] When the solid chip inspection device 1 starts processing (step S100), it makes a termination decision (step S102). The termination decision may be the end of the inspection, or it may be stopped manually. If it is to be terminated (Y branch of step S102), the device is stopped (step S104).

[0026] If the process continues (N branch of step S102), the process proceeds to step S106, where the vibration of the diaphragm 104 (or the auxiliary vibration member 16) connected to the test object 100 is measured while being excited (step S106). This process can be said to be a process of detecting the vibration of the solid chip while vibrating the diaphragm.

[0027] Although there is no particular limitation on the vibration method, it is preferable to vibrate at a plurality of frequencies. This is because by measuring responses at various frequencies, it is possible to detect defects of different shapes. For example, it is possible to continuously apply frequencies from several Hz to several tens of kHz to the vibration material 104 etc. in a short period of time, and measure the vibration of the test object 100 at that time. In other words, it is possible to obtain time-axis data of the vibration.

[0028] Next, the measurement data is Fourier transformed (step S108). The measured vibration data is time-axis data, so it is converted to frequency-axis data. This is because there is a conversion to a resonant frequency depending on the presence or absence of a defect. This can be said to be a process of converting vibration into a frequency signal.

[0029] Then, the frequency data is subjected to principal component analysis (step S110). Since the frequency data is sampled and output as digital data, the principal component analysis can be performed directly. The principal component analysis can be performed by a known method. Here, the first principal component (1stIG) and the second principal component (2ndIG) are extracted. This can be said to be a principal component analysis step in which the frequency signal is subjected to principal component analysis to calculate the first principal component and the second principal component of the solid chip.

[0030] Then, the products are judged to be normal or abnormal using the principal component plot (step S112). The principal component plot is made by measuring the first and second principal components of a number of normal products and a number of abnormal products that have been intentionally given defects, and plotting them on a two-dimensional coordinate system of the first principal component - the second principal component.

[0031] Figure 4 shows an example of a principal component plot. The horizontal axis is the first principal component, and the vertical axis is the second principal component. The first and second principal components are determined from multiple normal products (black circles) and abnormal products (white squares) and plotted on a two-dimensional plane. Here, the range where normal products are concentrated and the area where abnormal products are concentrated are separated. A boundary BD is drawn between these two, and the range from this boundary BD where normal products are concentrated is called the "normal product range", and the range where abnormal products are concentrated is called the "abnormal product range".

[0032] There is no particular limitation on how to draw the boundary line. It can be drawn artificially by a human, or it can be drawn by connecting points equidistant from the coordinates of normal and abnormal products. In addition, it is also possible to combine a support vector machine, DBSCAN clustering, or the like, which are well-known machine learning methods.

[0033] The abnormal products used in creating the principal component plot are normal products that have been artificially damaged, or samples that have been found to be abnormal in past production.

[0034] After the principal component plot is created in Fig. 4, the first and second principal components of the inspected object 100 are shown as black triangles. Since the inspected object 100 belongs to a range where normal products are concentrated rather than the boundary, the inspected object is judged to be normal. Conversely, if it belongs to the abnormal product range, it is judged to be abnormal.

[0035] The above process can be said to be a judgment process for determining whether a solid chip is normal or abnormal by comparing a principal component plot showing the range in the first and second principal components where normal products exist and the range in which abnormal products exist with the first and second principal components of the solid chip.

[0036] Referring again to FIG. 3, once it has been determined whether the test subject is normal or abnormal, the result is displayed (step S114). After display, the result may be notified to other devices. Then, it is determined whether or not remeasurement is required (step S116). If remeasurement is required (Y branch at step S116), the sample is changed or left the same, and processing returns to step S106. If remeasurement is not required (N branch at step S116), processing returns to step S102 and an end determination is made.

[0037] In this example, the normality or abnormality of the test object is judged by performing principal component analysis after Fourier transform, but other methods may be used as long as they are a method for judging whether the test object is normal or abnormal from the frequency characteristics obtained by Fourier transform. For example, there are cases where the judgment is made only from the frequency characteristics after Fourier transform. Depending on the type of solid chip, almost the same defects may occur, and in cases where it is sufficient to inspect such defects, it may be possible to judge only from the increase or decrease in the value of a specific frequency in the frequency characteristics after Fourier transform.

[0038] Therefore, the part enclosed by the dotted line in FIG. 3 can be said to be a process (step S111) for determining whether the solid-state chip is normal or abnormal based on the frequency signal.

[0039] Fig. 5 shows a process flow for creating the principal component plot exemplified in Fig. 4. When the process starts (step S200), a termination decision is made (step S202). The termination decision may be the end of the inspection, or it may be that the inspection is stopped manually. If the inspection is to be terminated (Y branch in step S202), the device is stopped (step S204).

[0040] If the process is to be continued (N branch in step S202), the process proceeds to step S206, where the specifications of the inspected object 100 are input. In this case, these include the sample number, whether the product is normal or abnormal, and the like.

[0041] Next, the vibration of the diaphragm 104 (or the auxiliary vibration member 16) connected to the test object 100 is measured while being vibrated (step S208).

[0042] Next, the measurement data is Fourier transformed (step S210), and the frequency data is subjected to principal component analysis (step S212) to extract the first and second principal components.

[0043] The values ​​are then plotted on the principal component plot (step S214). Needless to say, different marks are used depending on whether the sample is normal or abnormal. Next, a decision is made as to whether to measure the next sample (step S216), and if so (Y branch of step S216), the process proceeds to step S206. If not (N branch of step S216), the process proceeds to step S218.

[0044] In step S218, the boundary between the normal product range and the abnormal product range is set (step S218). The boundary is determined as described above.

[0045] In the solid-state chip inspection apparatus according to the present invention, the method for inspecting solid-state chips is executed by the controller 14. Since this is executed by software, it can be said that the controller 14 includes a "section" related to the steps of the inspection method.

[0046] Figure 6 shows the configuration of the controller 14. The excitation / measurement unit 40, which controls the vibration exciter 10 and vibration measurement instrument 12 shown in Figures 1 and 2, sends control signals Cv and Cm to the vibration exciter 10 and vibration measurement instrument 12 to apply vibration of a predetermined frequency, and obtains the measured value at that time by a notification signal Sm. Therefore, the excitation / measurement unit 40 outputs a time axis signal TD.

[0047] The Fourier transform unit 42 receives the time axis signal TD and converts it into a frequency signal FD. Therefore, the output of the Fourier transform unit 42 is the frequency signal FD.

[0048] The principal component analysis unit 44, which receives the frequency signal FD, performs principal component analysis on the frequency signal, and outputs a first principal component 1stIG and a second principal component 2ndIG. The genuine / fault determination unit 46, which receives this, compares it with a genuine / fault plot (principal component plot; the boundary BD may be written in) created in advance and determines whether the product is normal or abnormal. Therefore, the output of the genuine / fault determination unit 46 is a normal product (black circle) or an abnormal product (white square). The result display unit 48 transmits it from the controller 14 to the outside.

[0049] The controller 14 is also provided with a principal component plot creation unit 60. Here, there is a data acquisition unit 50 that acquires data by measuring a plurality of normal and abnormal products. The data acquisition unit 50 repeatedly uses the excitation / measurement unit 40, the Fourier transform unit 42, and the principal component analysis unit 44. The data acquisition unit 50 outputs a principal component plot (IG-PLOT). A boundary determination unit 62 that receives the IG-PLOT assigns boundaries to the principal component plot. Then, the principal component plot is completed.

[0050] Incidentally, since it is possible to determine whether the new test object 100 is in the normal range or the abnormal range even without a boundary, the boundary determining section 62 may not be provided. EXAMPLES

[0051] <Sample> A 500μm×400μm×100μm Si chip was used as the solid chip. An electronic circuit was formed on the surface of this chip. This solid chip was fixed on a 20mm×20mm×100μm PVC film (vibration material 104) with a 50μm thick sticky adhesive layer.

[0052] In addition, solid chips that were judged to have no defects in appearance by observation under a stereomicroscope were prepared as normal products, and those in which a part of a normal product had been scraped off to intentionally cause a defect were prepared as abnormal products.

[0053] <Vibrator> The exciter used was a laser diode pumped Nd:YAG laser. The laser diode pumped Nd:YAG laser can change the irradiation frequency f by changing the frequency of the applied current. The laser wavelength was 532 nm and the output was 25 mW.

[0054] The vibration pattern was a frequency range of 50 Hz to 400 Hz, with 1.4 Hz intervals for 20 ms each. Therefore, it took 5 seconds to apply all frequencies.

[0055] <Vibration measuring device> The vibration measurement was performed using a laser Doppler vibrometer. The vibration measurement results were input to a computer (controller 14), where they were Fourier transformed and then subjected to principal component analysis.

[0056] Figures 7 and 8 show schematic diagrams of the experiment and Fourier transformed waveforms. Figure 7(a) shows the case where an excitation laser LL is directly irradiated onto the test object 100, and the vibration of the test object 100 is measured with a laser Ls of a vibration measuring instrument. As shown in Figure 7(a-1), the vibration material 104 is measured while placed on a pedestal 20. Figure 7(a-2) shows the waveform after Fourier transformation of a normal product, and Figure 7(a-3) shows the waveform after Fourier transformation of an abnormal product. In both figures, the horizontal axis is frequency (Hz) and the vertical axis is intensity (unitless).

[0057] It seems possible to distinguish between normal and abnormal products based on the difference in intensity. However, this method involves directly irradiating the test object 100 with the excitation laser, which may damage the test object 100.

[0058] Fig. 7(b) shows the case where an excitation laser LL is irradiated onto the vibrating body 102 attached to the test object 100. Fig. 8(c) and Fig. 8(d) show the frequency characteristics after Fourier transform when an excitation laser LL is irradiated onto a vibration support material 16 made of a plastic case and when an excitation laser LL is irradiated onto a vibration support material 16 made of a metal plate, respectively.

[0059] Compared to the case of Figure 7(b) in which the vibration material 104 is made of flexible PET, the vibration frequency is higher, and in Figure 8(d) in which the auxiliary vibration material 16 is made of a metal plate, high-frequency components above 100 Hz are also clearly evident.

[0060] Figures 9 and 10 show principal component plots of the Fourier transformed data with the first and second principal components as axes, after principal component analysis. Figure 9(a) shows the results when the vibration member 104 in Figure 7(b) is vibrated, Figure 9(b) shows the results when the plastic case in Figure 8(a) is used as the vibration support member 16 and the vibration support member 16 is vibrated, and Figure 10 shows the results when the metal plate in Figure 8(b) is used as the vibration support member 16 and the vibration support member 16 is vibrated. Black circles indicate normal products, and white squares indicate abnormal products.

[0061] As can be seen by referring to Figures 9 and 10, there is a clear gap between normal and abnormal products, and the boundary can be easily drawn. The straight line BD in the figures is the boundary drawn by human judgment. When measuring an object 100 to be inspected, whose normality is unknown, the object 100 can be judged as normal or abnormal depending on whether it falls in the range to the right or left of this boundary.

[0062] Fig. 11 shows the principal component plots for a solid chip of indium phosphide (InP) measuring 500 μm × 600 μm × 100 μm. In Fig. 11(a), an aluminum plate is placed as a vibration support material 16 under the vibration material 104, and the vibration support material 16 is irradiated with a vibration laser LL. In Fig. 11(b), the vibration support material 16 is present, but the vibration laser LL is irradiated to the test object 100.

[0063] In both plots, the horizontal axis is the first principal component and the vertical axis is the second principal component. Referring to Fig. 11(a), when the vibration support material was excited, a clear boundary between normal and abnormal products could be created, but when the solid chip was excited directly, a partial overlapping area (circled) was created between normal and abnormal products, and normal and abnormal products could not be separated. [Industrial Applicability]

[0064] As described above, the solid-state chip inspection method according to the present invention can detect defects such as chips and scratches on small components such as semiconductor chips, which are difficult to visually inspect, without causing damage to the solid-state chip itself. [Explanation of symbols]

[0065] 1 Solid chip inspection equipment 2. Solid chip inspection equipment 10. Vibrator 12 Vibration measuring instrument 14 Controller 16 Vibration support material 40 Vibration and measurement section 42 Fourier transform section 44 Principal component analysis section 46 Authenticity determination section 48 Result display section 50 Data Acquisition Section 60 Plot Creation Department 62 Boundary Determination Section 100 Test subject 102 Adhesive layer 104 Vibration material

Claims

1. A method for inspecting a solid chip for detecting defects in the solid chip and judging whether the solid chip is normal or defective, comprising the steps of: Adhering the solid chip to a vibration material via an adhesive layer; detecting vibration of the solid-state tip while vibrating the vibration member; converting the vibration into a frequency signal; a step of determining whether the solid-state chip is normal or abnormal based on the frequency signal.

2. The step of determining whether the solid-state chip is a normal product or a defective product includes: a principal component analysis step of performing principal component analysis on the frequency signal to calculate a first principal component and a second principal component of the solid chip; 2. The solid-state chip inspection method according to claim 1, further comprising a step of comparing a principal component plot showing a range in which the normal product exists and a range in which the abnormal product exists in the first principal component and the second principal component with the first principal component and the second principal component of the solid-state chip, to determine whether the solid-state chip is normal or abnormal.

3. 3. A method for testing a solid chip according to claim 1, wherein the vibration material is made of two kinds of vibration materials stacked together.

Images