Analysis device and analysis method

Abstract

An analysis device 1 comprises: an LSI tester 3 that inputs a test pattern signal TP to a semiconductor device D and outputs a trigger signal indicating start timing of the test pattern signal TP and a test result signal TR indicating a test result, either Pass or Fail, of the semiconductor device D; an SSPD 13 that detects light emitted from a field of view set on the semiconductor device D and outputs a detection signal; and a processing unit 19 that processes the trigger signal and the detection signal to measure elapsed time from the start timing to the timing of the detection signal, generates light emission timing data indicating the elapsed time, classifies the light emission timing data into either Pass or Fail on the basis of the test result signal TR that is output from the LSI tester 3, and calculates, for each Pass and Fail, a total of frequency of the elapsed time indicating multiple light emission timing data when the test pattern signal TP is repeatedly input.

Description

Analysis Device and Analysis Method

[0001] One aspect of the embodiment relates to an analysis device and an analysis method.

[0002] Conventionally, techniques for performing emission analysis of semiconductor devices have been known. For example, Patent Document 1 below discloses a device that detects photons generated from an integrated circuit in accordance with a test pattern signal applied from a tester and performs data analysis processing based on the time when the photons are detected.

[0003] U.S. Patent Publication No. 2004 / 0189335

[0004] Franco Stellari et al., “Time-Resolved Imaging of VLSI Circuits using a Single-Point Single-Photon Detector and a Scanning Head”, ISTFA 2019: Conference Proceedings from the 45th International Symposium for Testing and Failure Analysis November 10-November 14, 2019.Franco Stellari et al., “1D and 2D Time-Resolved Emission Measurements of Circuits Fabricated in 14 nm Technology Node”, 2020 IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA), 20-23 July 2020.

[0005] In the conventional device as described above, emission analysis according to the operating state of the semiconductor device has not been performed. Therefore, it is desired to achieve more efficient failure analysis for semiconductor devices.

[0006] Therefore, one aspect of the embodiment was made in view of the above problems, and aims to provide an analysis apparatus and analysis method that realize efficient failure analysis of semiconductor devices.

[0007] The analysis apparatus according to the first aspect of the embodiment includes: a tester that inputs a test pattern to a semiconductor device and outputs a trigger signal indicating the start timing of the test pattern and a test result signal indicating Pass or Fail, which is the test result of the semiconductor device based on the test pattern; a superconducting single-photon detector that detects light generated from a field of view set on the semiconductor device in response to the input of the test pattern and outputs a detection signal; and a processing unit that processes the trigger signal and the detection signal to measure the elapsed time from the start timing to the timing of the detection signal, generates light emission timing information indicating the elapsed time, classifies the light emission timing information into either Pass or Fail based on the test result signal output from the tester, and aggregates the frequency of the elapsed time indicated by the multiple light emission timing information when the tester repeatedly inputs the test pattern for each Pass and Fail.

[0008] Alternatively, the analysis method relating to a second aspect of the embodiment includes the steps of: using a tester to input a test pattern to a semiconductor device and outputting a trigger signal indicating the start timing of the test pattern and a test result signal indicating Pass or Fail, which is the test result of the semiconductor device based on the test pattern; using a superconducting single-photon detector to detect light generated from a field of view set on the semiconductor device in response to the input of the test pattern and outputting a detection signal; using a processing unit to process the trigger signal and the detection signal to measure the elapsed time from the start timing to the timing of the detection signal and generate light emission timing information indicating the elapsed time; using a processing unit to classify the light emission timing information into either Pass or Fail based on the test result signal output from the tester; and using a processing unit to aggregate the frequency of the elapsed time indicated by the multiple light emission timing information when the tester repeatedly inputs the test pattern, for each Pass and Fail.

[0009] According to the first or second aspect described above, the test results of a semiconductor device into which a test pattern has been input are output by the tester. The elapsed time from the start of the test pattern to the detection of light generated from the field of view on the semiconductor device is measured, and the frequency of the elapsed time when the test pattern is input is aggregated for each Pass and Fail test result. This makes it possible to analyze the trend of light emission timing in the field of view for each operating state of the semiconductor device corresponding to the test pattern, and to efficiently perform failure analysis of the semiconductor device.

[0010] Any aspect of this disclosure can enable efficient failure analysis of semiconductor devices.

[0011] Figure 1 is a schematic diagram of the analysis apparatus 1 according to the embodiment. Figure 2 is a diagram showing the scanning state of the field of view on the semiconductor device D by the field of view scanning unit 5 of Figure 1. Figure 3 is a timing chart showing the classification processing by the PC 17. Figure 4 is a graph showing an image of the aggregated data generated by the PC 17. Figure 5 is a graph showing the aggregated data actually generated by the PC 17. Figure 6 is a “Pass” emission image G output by the PC 17. Ｐ and “Fail” emission image G Ｆ This figure shows an example. Figure 7 is a flowchart showing the procedure for failure analysis using the analysis device 1. Figure 8 is a schematic configuration diagram of the analysis device 1A according to the first modified example. Figure 9 is a schematic configuration diagram of the analysis device 1B according to the second modified example. Figure 10 is a schematic configuration diagram of the analysis device 1C according to the third modified example. Figure 11 is a diagram showing images of the emission waveform and detected waveform in the field of view. Figure 12 is a diagram showing an image of the composite waveform shown by aggregated data generated from emission timing data of the first to fourth wavelength components in the third modified example. Figure 13 is a diagram showing the types of aggregated data that can be generated in the third modified example.

[0012] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In this description, the same reference numerals will be used for elements that are the same or have the same function, and redundant explanations will be omitted.

[0013] Figure 1 is a schematic diagram of the analysis apparatus 1 according to the embodiment. The analysis apparatus 1 is a device for performing failure analysis on a semiconductor device that is a device under test (DUT). Examples of semiconductor devices include integrated circuits having PN junctions such as transistors (e.g., small-scale integrated circuits (SSI), medium-scale integrated circuits (MSI), large-scale integrated circuits (LSI), very large-scale integrated circuits (VLSI), ultra-large-scale integrated circuits (ULSI), giga-scale integrated circuits (GSI)), high-current / high-voltage MOS transistors and bipolar transistors, power semiconductor elements (power devices), etc. In addition, the semiconductor device may be a package containing the semiconductor device, a composite substrate, etc.

[0014] As shown in Figure 1, the analysis device 1 comprises an LSI tester 3, an optical device 11 including a field-of-view scanning unit (scanning section) 5, a revolving nosepiece 7, and an objective lens unit 9, a superconducting single-photon detector (SSPD) 13, and a processing unit 19 including a time measuring instrument 15 and a personal computer (PC) 17. The configuration of each component of the analysis device 1 will be described below.

[0015] The LSI tester 3 is electrically connected to the semiconductor device D, which is the DUT (Device Under Test). The LSI tester 3 inputs a test pattern signal TP (test pattern) as an input signal to the semiconductor device D. The test pattern signal TP is an analytical signal having a certain pattern. The LSI tester 3 repeatedly inputs the test pattern signal TP to the semiconductor device D at a fixed period. The test pattern signal TP is set, for example, by the user. In addition, each time the test pattern signal TP is input, the LSI tester 3 performs a test based on the output signal output from the semiconductor device D and generates a test result signal TR indicating Pass (success) or Fail (failure). That is, the LSI tester 3 compares the output signal output from the semiconductor device D with the design-correct output signal for the test pattern signal TP and outputs a test result signal TR based on the comparison result (test result). For example, LSI tester 3 outputs a test result signal TR indicating "Pass" if the output signal matches the correct signal, and outputs a test result signal TR indicating "Fail" if the output signal does not match the correct signal. In addition, LSI tester 3 outputs a trigger signal indicating the start timing of the test pattern signal TP for each input of the test pattern signal TP.

[0016] In LSI Tester 3, the operating state of semiconductor device D (e.g., operating voltage, timing conditions, clock frequency, etc.) is pre-set so that the test result for semiconductor device D is either Pass or Fail with a 50% probability.

[0017] In the semiconductor device D, extremely weak light (photons) is generated by operation based on the test pattern signal TP. The optical device 11 is a device that guides the light generated in the field of view set on the semiconductor device D toward the SSPD 13. The optical device 11 consists of a field-of-view scanning unit 5, a revolving nosepiece 7, and an objective lens unit 9. The objective lens unit 9 includes multiple objective lenses 9a, 9b, and 9c with different magnifications that guide the light generated from the semiconductor device D to the field-of-view scanning unit 5. Note that the number of objective lenses included in the objective lens unit 9 is not limited to three, but can include any number of objective lenses, such as five to ten. The revolving nosepiece 7 supports the multiple objective lenses 9a, 9b, and 9c included in the objective lens unit 9 in a switchable manner on the optical path between the semiconductor device D and the field-of-view scanning unit 5, and guides the light generated in the semiconductor device D to the field-of-view scanning unit 5 via any of the objective lenses 9a, 9b, and 9c. The field of view scanning unit 5 is optically connected to the SSPD 13 by an optical fiber F1, and inputs light guided by the objective lens unit 9 to the SSPD 13. The field of view scanning unit 5 incorporates an optical scanner such as a galvanometer mirror, and scans the field of view on the semiconductor device D in two dimensions, inputting the light from the field of view set by the scan to the SSPD 13. The field of view scanning unit 5 also outputs position information indicating the two-dimensional position of the field of view on the semiconductor device D set by the scan to the PC 19, which will be described later. Figure 2 shows the scanning state of the field of view on the semiconductor device D by the field of view scanning unit 5. In this way, the field of view scanning unit 5 moves the field of view VF along the X-axis direction, which is one direction along the surface of the semiconductor device D, and the movement of the field of view VF along the X-axis direction is repeated while being shifted in the Y-axis direction perpendicular to the X-axis direction along the surface of the semiconductor device D, so that the field of view VF is repeatedly set while being shifted over the entire surface of the semiconductor device D. The field of view scanning unit 5 can also receive light (such as a laser) from a light source (not shown) and irradiate the field of view position on the semiconductor device D with light.However, since the SSPD 13 may be damaged by strong light such as a laser, when the field of view scanning unit 5 irradiates light onto the semiconductor device D, it is necessary to configure the system so that reflected light from the semiconductor device D does not input to the SSPD 13.

[0018] The SSPD 13 is a point sensor that utilizes a superconducting material to detect single photons within a single field of view, and is also known as an SNSPD (Superconducting Nanowire Single Photon Detector). The SSPD 13 detects light generated from a field of view set on the semiconductor device D via the optical device 11, and accordingly outputs a detection signal, which is an electrical pulse synchronized with the timing of the light detection.

[0019] The processing unit 19 is composed of a time measuring instrument 15 and a PC 17, and is a signal processing unit that processes various signals from the SSPD 13 and the LSI tester 3. The time measuring instrument 15 is electrically connected to the LSI tester 3 and the SSPD 13. The PC 17 is electrically connected to the LSI tester 3, the field of view scanning unit 5, and the time measuring instrument 15.

[0020] The time measuring device 15 processes the trigger signal input from the LSI tester and the detection signal output from the SSPD 13 for each field of view set on the semiconductor device D to generate light emission timing data (light emission timing information). The time measuring device 15 can be a TAC (Time-to-Analog Converter), a TDC (Time-to-Digital Converter), or a digitizer. In other words, the time measuring device 15 measures the elapsed time from the start timing indicated by the trigger signal output in response to the test pattern signal TP input to the semiconductor device D to the light detection timing indicated by the detection signal output from the SSPD 13, and generates light emission timing data indicating the elapsed time. The time measuring device 15 repeatedly generates multiple light emission timing data for each periodically repeated test pattern signal TP, and for each of the multiple lights (photons) detected after the input of the test pattern signal TP, and sequentially outputs the multiple light emission timing data to the PC 17.

[0021] PC 17 classifies the light emission timing data output by the time measuring instrument 15 into either Pass or Fail for each field of view set on the semiconductor device D. Specifically, PC 17 examines each light emission timing data and identifies the test result signal TR output in the period that includes the timing of light generation. If the test result signal TR indicates Pass, it classifies it as Pass and attaches the classification flag "Pass" to the light emission timing data. If the test result signal TR indicates Fail, it classifies it as Fail and attaches the classification flag "Fail" to the light emission timing data.

[0022] Figure 3 is a timing chart showing the classification process by the PC 17. While one field of view is set by the field of view scanning unit 5, a test pattern signal TP is repeatedly input to the semiconductor device D from the LSI tester 3 at a fixed interval. Multiple light emission timing data are generated by the time measuring instrument 15 in response to multiple lights generated in one cycle, from the input of one test pattern signal TP to the input of the next test pattern signal TP. The classification flag "Pass" indicated by the test result signal TR output by the LSI tester 3 in that cycle is assigned to these multiple light emission timing data. Similarly, the classification flag "Fail" indicated by the test result signal TR output by the LSI tester 3 in that cycle is assigned to the multiple light emission timing data generated in the next cycle. Note that the data acquisition method is not limited to this; after inputting the test pattern signal TP multiple times and collecting data in a field of view, the field of view may be moved, and the process of inputting the test pattern signal TP multiple times in the next field of view and collecting data in that field of view may be repeated. Alternatively, after inputting the test pattern signal TP once and collecting data in a given field of view, the field of view may be moved, the test pattern signal TP may be input once in the next field of view, the data for that field of view may be repeated, and then the system may return to the initial field of view position and repeat the data collection process.

[0023] In addition, PC17 aggregates the classified multiple light emission timing data for each field of view set on the semiconductor device D to generate aggregated data on the frequency of elapsed time. For example, PC17 aggregates the frequency of elapsed time indicated by multiple light emission timing data classified as "Pass" by dividing them into bins of a predetermined time width to generate "Pass" aggregated data. PC17 also aggregates the frequency of elapsed time indicated by multiple light emission timing data classified as "Fail" by dividing them into bins of a predetermined time width to generate "Fail" aggregated data. Furthermore, PC17 subtracts the frequency of elapsed time indicated by multiple light emission timing data classified as "Pass" from the frequency of elapsed time indicated by multiple light emission timing data classified as "Fail" for each of the divided elapsed time bins to generate differential aggregated data.

[0024] Figure 4 is a graph showing an image of the aggregated data generated by PC17, and Figure 5 is a graph showing the aggregated data actually generated by PC17. Figure 4(a) shows aggregated data generated using all emission timing data for one field of view, Figure 4(b) shows aggregated data generated using "Pass" emission timing data for one field of view, and Figure 4(c) shows aggregated data generated using "Fail" emission timing data for one field of view. In this way, the "Pass" aggregated data and the "Fail" aggregated data can be used to clarify the difference in the trend of light generation timing from one field of view based on the input of the test pattern signal TP between cases where the test result is "Pass" and cases where the test result is "Fail". Figure 5 shows the "Pass" aggregated data and "Fail" aggregated data generated by PC17 when the input period of the test pattern signal TP is set to 20us. These examples of aggregated data show that the trends in aggregated values ​​(frequency, count) around an elapsed time of 2.32 us are the same, but the trends in aggregated values ​​around an elapsed time of 2.46 us are different, indicating that when the test result is "Pass", the frequency of light emission is low, and when the test result is "Fail", the frequency of light emission is high.

[0025] Furthermore, the PC 17 uses the "Pass" aggregate data generated for each field of view and the position information regarding the two-dimensional position of each field of view output from the field of view scanning unit 5 to create a "Pass" emission image for a time width centered on a desired elapsed time. Specifically, the PC 17 assigns each field of view on the semiconductor device D indicated by the position information to a pixel, and sets the integrated value of the bin frequencies included in the above time width in the "Pass" aggregate data for each field of view as the pixel value of each pixel, thereby creating a "Pass" emission image for a time width centered on a desired elapsed time. Similarly, the PC 17 uses the "Fail" aggregate data generated for each field of view and the position information regarding the two-dimensional position of each field of view to create a "Fail" emission image for a time width centered on a desired elapsed time. In addition, the PC 17 uses the difference aggregate data generated for each field of view and the position information regarding the two-dimensional position of each field of view to create a difference emission image for a time width centered on a desired elapsed time. This difference image represents the difference in pixel values ​​between the "Pass" image and the "Fail" image. The PC 17 then outputs the created "Pass" image, "Fail" image, and difference image to an output device such as a display, a data storage device, etc. provided in the PC 17. The PC 17 may output the three types of light-emitting images ("Pass" image, "Fail" image, and difference image) within a predetermined time interval as still images, or it may output the three types of light-emitting images ("Pass" image, "Fail" image, and difference image) within multiple consecutive time intervals as moving images.

[0026] Figure 6 shows the “Pass” emission image G output by PC17. Ｐ and “Fail” emission image G Ｆ This is an example shown in Figure 6. Figure 6 is an image created based on the aggregated data shown in Figure 5, showing a "Pass" emission image G over a time interval centered around an elapsed time of 2.32 us. Ｐ and “Fail” emission image G Ｆ And, “Pass” emission image G over a time window centered around the elapsed time of 2.46 us. Ｐ and “Fail” emission image G ＦThese are examples. These images show that the trend in light emission frequency around an elapsed time of 2.32 us is the same across the entire semiconductor device D, but the trend in light emission frequency around an elapsed time of 2.46 us is different. When the test result is "Pass", the light emission frequency is generally low, and when the test result is "Fail", the light emission frequency is high in some areas on the semiconductor device D.

[0027] Next, the procedure for failure analysis of semiconductor devices using the analysis device 1 described above will be explained, and the analysis method according to the embodiment will be described in detail. Figure 7 is a flowchart showing the procedure for failure analysis using the analysis device 1.

[0028] First, the initial field of view is set on the semiconductor device D using the field of view scanning unit 5 (step S1). Next, the LSI tester 3 repeatedly inputs a test pattern signal TP to the semiconductor device D at a predetermined interval, and accordingly, a trigger signal and a test result signal TR are output (step S2).

[0029] In parallel with the input of the test pattern signal TP, the SSPD 13 detects light from the field of view of the semiconductor device D and outputs a detection signal (step S3). Subsequently, the time measuring instrument 15 generates multiple light emission timing data indicating the elapsed time from the input of the test pattern signal TP to the detection timing of multiple light (photons) based on the trigger signal and the multiple detection signals (step S4).

[0030] Next, the PC 17 uses the test result signal TR output from the LSI tester 3 to classify multiple light emission timings, and generates three types of aggregated data: "Pass" aggregated data, "Fail" aggregated data, and differential aggregated data (step S5). Furthermore, the field of view scanning unit 5 determines whether or not to scan and change the field of view on the semiconductor device D (step S6). If scanning is to be performed (step S6; Yes), the field of view scanning unit 5 sets the next field of view on the semiconductor device D, and the processing in steps S1 to S5 is repeated, thereby generating three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and differential aggregated data) for each field of view on the semiconductor device D.

[0031] Finally, the PC 17 processes the three types of aggregated data for each field of view, creating and outputting three types of emission images for the desired elapsed time: a "Pass" emission image, a "Fail" emission image, and a difference emission image (step S7). This completes the failure analysis process for semiconductor device D.

[0032] The effects and advantages of this embodiment will now be explained.

[0033] In the analysis apparatus 1 and analysis method using the same according to this embodiment, the test results of the semiconductor device D to which the test pattern signal TP is input are output by the LSI tester 3. The elapsed time from the start timing of the test pattern signal TP to the detection timing of light generated from the field of view on the semiconductor device D is measured, and the frequency of the elapsed time when the input of the test pattern signal TP is repeated is aggregated for each Pass and Fail test result. This makes it possible to analyze the trend of light emission timing in the field of view for each operating state of the semiconductor device D corresponding to the test pattern signal TP, and to efficiently perform failure analysis of the semiconductor device.

[0034] In this embodiment, by generating differential aggregate data, it is possible to analyze the difference in the trend of light emission timing in the field of view on the semiconductor device D between when the test result is Pass and when the test result is Fail, thereby enabling more efficient failure analysis of the semiconductor device.

[0035] Furthermore, in this embodiment, a field-of-view scanning unit 5 is used to scan the field of view two-dimensionally on the semiconductor device D. In this case, the trend of light emission timing in a wide field of view of the semiconductor device D can be analyzed for each operating state of the semiconductor device D corresponding to the test pattern signal TP, and failure analysis of the entire semiconductor device can be performed efficiently.

[0036] Furthermore, in this embodiment, "Pass" and "Fail" emission images are created at a desired elapsed time. In this case, the trend of emission timing across a wide field of view of the semiconductor device D can be visually analyzed for each operating state of the semiconductor device D corresponding to the test pattern signal TP, enabling more efficient failure analysis of the entire semiconductor device.

[0037] Furthermore, in this embodiment, a differential emission image is created. In this case, the difference in the trend of emission timing across a wide field of view on the semiconductor device D between when the test result is Pass and when the test result is Fail can be visually analyzed, enabling more efficient failure analysis of the entire semiconductor device.

[0038] Conventionally, DALS (Dynamic Analysis by Laser Stimulation), which is one of the dynamic semiconductor failure analysis methods, is known. DALS is a method of setting a voltage or frequency at the boundary of the operating state of a semiconductor device, changing the resistance or transistor characteristics by the irradiation heat of a laser, and finding abnormalities in the circuit inside the semiconductor device, the critical path of a signal, etc. In particular, in DALS, when irradiating with a laser, it is determined whether the output from the semiconductor device is in a Pass state or a Fail state using a tester, and the operating state such as the voltage or frequency that becomes the boundary between the Pass state and the Fail state is analyzed.

[0039] In recent years, with the multilayerization of semiconductor devices, it has become increasingly difficult to irradiate laser light onto the circuit part inside the semiconductor device. Also, the amount of emission light generated in the circuit part inside the semiconductor device and output to the outside of the semiconductor device tends to be weak. Even under such circumstances, according to the present embodiment, a test pattern signal TP with a voltage or clock frequency corresponding to the operating boundary is periodically input to the semiconductor device D, weak emission light from the semiconductor device D accompanying the test pattern signal TP is detected, and analysis of the emission timing is performed. In the analysis of the emission timing, the emission timing is classified according to whether the test result of the semiconductor device D is Pass or Fail, and the tendency of the emission timing is analyzed. Thereby, even when the semiconductor device is multilayered, it is possible to efficiently perform failure analysis of the entire semiconductor device. That is, it is possible to identify the emission timing peculiar to the Pass state and the emission timing peculiar to the Fail state.

[0040] As described above, various embodiments of the present invention have been explained, but the present invention is not limited to the above embodiments, and may be modified without changing the gist described in each claim, or may be applied to other things.

[0041] FIG. 8 is a schematic configuration diagram of an analysis device 1A according to the first modification example. In the analysis device 1A, the functions and connection configurations of the time measurement device 15A and the PC 17A are different from those of the above-described embodiment.

[0042] That is, the time measuring device 15A is electrically connected to the LSI tester 3, and in addition to the trigger signal, a test result signal TR is also input from the LSI tester 3. The time measuring device 15A has a function of generating classification flags for a plurality of emission timing data for each field of view in addition to the function of generating a plurality of emission timing data for each field of view. The time measuring device 15A classifies a plurality of emission timing data based on the test result signal TR, and attaches a classification flag to each of the plurality of emission timing data. The PC 17A acquires a plurality of emission timing data with classification flags from the time measuring device 15A for each field of view, and generates three types of aggregated data for each field of view (“Pass” aggregated data, “Fail” aggregated data, and difference aggregated data). In addition, the PC 17A creates and outputs three types of emission images (“Pass” emission image, “Fail” emission image, and difference image) based on the three types of aggregated data for each field of view and the position information for each field of view input from the field scan unit 5.

[0043] According to the first modification example, the classification process of the emission timing data is made more efficient, and the analysis process is speeded up.

[0044] FIG. 9 is a schematic configuration diagram of an analysis device 1B according to a second modification example. In the analysis device 1B, the points that two SSPDs 13Ba and 13Bb and two time measuring devices 15Ba and 15Bb are provided, and the function of the PC 17B is different from that of the above-described embodiment.

[0045] The two SSPDs 13Ba and 13Bb are optically connected to the output of the field-of-view scanning unit 5 via a polarization separation unit F2. The polarization separation unit F2 separates the transversely polarized light, which is the first polarization, from the light pulse output from the field-of-view scanning unit 5 and guides it to SSPD 13Ba, and separates the longitudinally polarized light, which is the second polarization, and guides it to SSPD 13Bb. Such a polarization separation unit F2 is a polarization separation fiber that separates light according to its polarization. The polarization separation unit F2 may consist of a branched fiber that does not have polarization separation characteristics and two polarizing plates provided on the front surfaces of SSPDs 13Ba and Bb, respectively, or it may consist of a branched fiber and a polarization beam splitter arranged at the branching portion of the branched fiber. SSPD 13Ba detects transversely polarized light generated from the field of view on the semiconductor device D and outputs a detection signal. SSPD 13Bb detects longitudinally polarized light generated from the field of view on the semiconductor device D and outputs a detection signal.

[0046] The time measuring device 15Ba is electrically connected to the SSPD 13Ba and the LSI tester 3, and generates multiple emission timing data for transversely polarized light for each field of view. The time measuring device 15Bb is electrically connected to the SSPD 13Bb and the LSI tester 3, and generates multiple emission timing data for longitudinally polarized light for each field of view.

[0047] PC17B generates three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for each field of view, based on multiple emission timing data of transverse polarization generated by the time measuring instrument 15Ba. PC17B also generates three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for each field of view, based on multiple emission timing data of longitudinal polarization generated by the time measuring instrument 15Bb. Furthermore, PC17B generates three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for each field of view, based on multiple emission timing data of two polarizations generated by the two time measuring instruments 15Ba and 15Bb. Finally, PC17B creates and outputs an emission image using the aggregated data generated for each field of view.

[0048] According to the second modified example, the trend of emission timing of predetermined polarizations in the field of view can be analyzed for each operating state of the semiconductor device D corresponding to the test pattern signal TP. The polarization state of the light emitted to the outside of the semiconductor device D differs depending on the internal structure of the semiconductor device D. By generating an emission image for each polarization state, failure analysis can be efficiently performed for each internal circuit structure of the semiconductor device D.

[0049] Figure 10 is a schematic diagram of the analysis device 1C according to the third modified example. The analysis device 1C differs from the embodiment described above in that it is equipped with four SSPDs 13Ca, 13Cb, 13Cc, and 13Cd, and four time measuring instruments 15Ca, 15Cb, 15Cc, and 15Cd, and the function of the PC 17C is different.

[0050] The four SSPDs 13Ca, 13Cb, 13Cc, and 13Cd are optically connected to the output of the field-of-view scanning unit 5 via a wavelength separation unit F3. The wavelength separation unit F3 separates the light of a first wavelength component (e.g., 1100 nm to 1500 nm) from the light pulse output from the field-of-view scanning unit 5 and leads it to SSPD 13Ca, separates the light of a second wavelength component (e.g., 1500 nm to 1800 nm) and leads it to SSPD 13Cb, separates the light of a third wavelength component (e.g., 1800 nm to 2000 nm) and leads it to SSPD 13Cc, and separates the light of a fourth wavelength component (e.g., 2000 nm to 2300 nm) and leads it to SSPD 13Cd. Such a wavelength separation unit F3 is an optical fiber coupler that separates light according to its wavelength. The wavelength separation unit F3 may also be an optical element such as a dichroic mirror or a diffraction grating. In this case, the wavelength separation unit F3 may output a spatial output, separate it using an optical element such as a dichroic mirror or diffraction grating, and then recouple it to a fiber for input to each of the SSPDs 13Ca to 13Cd. SSPD 13Ca detects light of a first wavelength component originating from a certain field of view on the semiconductor device D and outputs a detection signal. SSPD 13Cb detects light of a second wavelength component originating from the same field of view on the semiconductor device D and outputs a detection signal. SSPD 13Cc detects light of a third wavelength component originating from the same field of view on the semiconductor device D and outputs a detection signal. SSPD 13Cd detects light of a fourth wavelength component originating from the same field of view on the semiconductor device D and outputs a detection signal.

[0051] Time measuring device 15Ca is electrically connected to SSPD 13Ca and LSI tester 3, and generates multiple emission timing data for the first wavelength component of light for each field of view. Time measuring device 15Cb is electrically connected to SSPD 13Cb and LSI tester 3, and generates multiple emission timing data for the second wavelength component of light for each field of view. Time measuring device 15Cc is electrically connected to SSPD 13Cc and LSI tester 3, and generates multiple emission timing data for the third wavelength component of light for each field of view. Time measuring device 15Cd is electrically connected to SSPD 13Cd and LSI tester 3, and generates multiple emission timing data for the fourth wavelength component of light for each field of view.

[0052] PC17C generates three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for each field of view, based on multiple emission timing data of the first wavelength component generated by the time measuring instrument 15Ca. PC17C also generates three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for each field of view, based on multiple emission timing data of the second wavelength component generated by the time measuring instrument 15Cb. Similarly, the PC17C generates, for each field of view, three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for the third wavelength component and three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) for the fourth wavelength component, based on multiple emission timing data of the third and fourth wavelength components generated by the two time measuring instruments 15Cc and 15Cd. Furthermore, the PC17C may generate, for each field of view, three types of aggregated data ("Pass" aggregated data, "Fail" aggregated data, and difference aggregated data) that take into account multiple wavelength components selected from the four wavelength components.

[0053] In the third modified example, the detection waveform analyzed by the detection signal can be corrected by generating aggregated data that takes into account multiple wavelength components. Figure 11 shows an image of the emission waveform and detection waveform in the field of view. As described above, the propagation characteristics of the optical path from the semiconductor device D to the SSPD are wavelength-dependent, and in particular the wavelength separation unit F3 including the optical fiber has wavelength dispersion characteristics. Therefore, the detection waveform analyzed with respect to the emission waveform includes the first to fourth wavelength components W1 to W4, and as a result of the first to fourth wavelength components W1 to W4 being shifted relative to each other in the time direction, the detection waveform expands in the time direction. When PC17C generates aggregated data targeting emission timing data of multiple wavelength components among the first to fourth wavelength components, it can correct the detection waveform to be closer to the emission waveform by aggregating the data while correcting the difference in propagation speed for each of the multiple wavelength components.

[0054] Figure 12 shows an image of a composite waveform represented by aggregated data generated from emission timing data of the first to fourth wavelength components in the third modified example. In this way, the waveform of "detection waveform 1" represented by multiple emission timing data of the first wavelength component, the waveform of "detection waveform 2" represented by multiple emission timing data of the second wavelength component, the waveform of "detection waveform 3" represented by multiple emission timing data of the third wavelength component, and the waveform of "detection waveform 4" represented by multiple emission timing data of the fourth wavelength component are combined while taking into account the difference in propagation delay time, thereby obtaining a composite waveform that approximates the emission waveform.

[0055] In the third modified example, aggregated data may be generated using emission timing data of any multiple wavelength components selected from the first to fourth wavelength components. Figure 13 shows the types of aggregated data that can be generated in the third modified example.

[0056] The emission timing data classified as "Pass," known as "P-time data," includes "P-time data (wavelength 1)," which is the emission timing data for the first wavelength component; "P-time data (wavelength 2)," which is the emission timing data for the second wavelength component; "P-time data (wavelength 3)," which is the emission timing data for the third wavelength component; and "P-time data (wavelength 4)," which is the emission timing data for the fourth wavelength component. The emission timing data classified as "F-time data," known as "F-time data (wavelength 1)," which is the emission timing data for the first wavelength component; "F-time data (wavelength 2)," which is the emission timing data for the second wavelength component; "F-time data (wavelength 3)," which is the emission timing data for the third wavelength component; and "F-time data (wavelength 4)," which is the emission timing data for the fourth wavelength component.

[0057] In the third modified example, in addition to "P aggregate data (wavelength 1)" which is "Pass" aggregate data targeting the first wavelength component, "P aggregate data (wavelength 2)" which is "Pass" aggregate data targeting the second wavelength component, "P aggregate data (wavelength 3)" which is "Pass" aggregate data targeting the third wavelength component, and "P aggregate data (wavelength 4)" which is "Pass" aggregate data targeting the fourth wavelength component, "P aggregate data (wavelength 1+2)" and "P aggregate data (wavelength 1+3)" which are "Pass" aggregate data synthesized from two wavelength components, "P aggregate data (wavelength 1+2+3)" which is "Pass" aggregate data synthesized from three wavelength components, and "P aggregate data (wavelength 1+2+3+4)" which is "Pass" aggregate data synthesized from four wavelength components may be generated. In addition, "F-aggregate data (wavelength 1)" is generated for the first wavelength component, "F-aggregate data (wavelength 2)" is generated for the second wavelength component, "F-aggregate data (wavelength 3)" is generated for the third wavelength component, and "F-aggregate data (wavelength 4)" is generated for the fourth wavelength component. Furthermore, "F-aggregate data (wavelength 1+2)" and "F-aggregate data (wavelength 1+3)" are generated by combining two wavelength components, "F-aggregate data (wavelength 1+2+3)" is generated by combining three wavelength components, and "F-aggregate data (wavelength 1+2+3+4)" is generated by combining four wavelength components.

[0058] According to the third modified example described above, it is possible to perform failure analysis of a semiconductor device by canceling out the influence of wavelength dispersion characteristics in the optical path from the field of view of the semiconductor device to the SSPD.

[0059] In the above embodiment, it is preferable for the processing unit to subtract the frequency of elapsed time classified as "Pass" from the frequency of elapsed time classified as "Fail" for each elapsed time. With this configuration, it is possible to analyze the difference in the trend of light emission timing in the field of view on the semiconductor device between when the test result is "Pass" and when the test result is "Fail," and to perform failure analysis of the semiconductor device more efficiently.

[0060] In the above embodiment, it is also preferable to further include a scanning unit that scans the field of view two-dimensionally on the semiconductor device. In this case, it is possible to analyze the trend of light emission timing in a wide field of view of the semiconductor device for each operating state of the semiconductor device according to the test pattern, and to efficiently perform failure analysis of the entire semiconductor device.

[0061] Furthermore, in the above embodiment, it is also preferable for the processing unit to create a light emission image of a Pass at a desired elapsed time based on positional information relating to the field of view and the frequency of elapsed time in Pass, and to create a light emission image of a Fail at a desired elapsed time based on positional information relating to the field of view and the frequency of elapsed time in Fail. In this case, the trend of light emission timing in a wide field of view of the semiconductor device can be visually analyzed for each operating state of the semiconductor device according to the test pattern, and failure analysis of the entire semiconductor device can be performed more efficiently.

[0062] Furthermore, in the above embodiment, it is also preferable for the processing unit to create a difference image representing the difference between the light emission image for Pass and the light emission image for Fail. In this case, the difference in the trend of light emission timing over a wide field of view on the semiconductor device between when the test result is Pass and when the test result is Fail can be visually analyzed, and failure analysis of the entire semiconductor device can be performed more efficiently.

[0063] Furthermore, in the above embodiment, it is also preferable to further include a polarization separation unit that separates the light emitted from the semiconductor device according to its polarization based on the test pattern. This makes it possible to analyze the trend of emission timing of predetermined polarizations in the field of view for each operating state of the semiconductor device according to the test pattern, and to efficiently perform failure analysis for each circuit structure in the semiconductor device.

[0064] Furthermore, in the above embodiment, it is also preferable to further include a wavelength separation unit that separates the light generated from the semiconductor device according to the wavelength based on the test pattern. With such a configuration, it is possible to analyze the trend of emission timing in a predetermined wavelength range in the field of view for each operating state of the semiconductor device according to the test pattern, and to perform failure analysis of the semiconductor device by canceling the influence of the wavelength dispersion characteristics of the optical path of the light.

[0065] 1, 1A, 1B, 1C... Analysis device, 3... LSI tester, 5... Field of view scanning unit (scanning unit), 13... Superconducting single-photon detector (SSPD), 19... Processing unit, D... Semiconductor device, F2... Polarization separation unit, F3... Wavelength separation unit, TP... Test pattern signal, LP... Light (photon), TR... Test result signal, VF... Field of view.

Claims

1. An analysis device comprising: a tester that inputs a test pattern to a semiconductor device and outputs a trigger signal indicating the start timing of the test pattern and a test result signal indicating Pass or Fail, which is the test result of the semiconductor device based on the test pattern; a superconducting single-photon detector that detects light generated from a field of view set on the semiconductor device in response to the input of the test pattern and outputs a detection signal; a processing unit that processes the trigger signal and the detection signal to measure the elapsed time from the start timing to the timing of the detection signal, generates light emission timing information indicating the elapsed time, classifies the light emission timing information into either Pass or Fail based on the test result signal output from the tester, and aggregates the frequency of the elapsed time indicated by the multiple light emission timing information when the tester repeatedly inputs the test pattern, for each Pass and Fail.

2. The analysis apparatus according to claim 1, wherein the processing unit subtracts the frequency of the elapsed time classified as Pass from the frequency of the elapsed time classified as Fail for each elapsed time.

3. The analysis apparatus according to claim 1 or 2, further comprising a scanning unit for scanning the field of view in two dimensions on the semiconductor device.

4. The analysis apparatus according to claim 3, wherein the processing unit creates a light emission image of a Pass at a desired elapsed time based on positional information relating to the field of view and the frequency of the elapsed time in a Pass, and creates a light emission image of a Fail at a desired elapsed time based on positional information relating to the field of view and the frequency of the elapsed time in a Fail.

5. The analysis apparatus according to claim 4, wherein the processing unit creates a difference image representing the difference between the light emission image of Pass and the light emission image of Fail.

6. The analysis apparatus according to any one of claims 1 to 5, further comprising a polarization separation unit for separating the light generated from the semiconductor device by the test pattern according to its polarization.

7. The analysis apparatus according to any one of claims 1 to 5, further comprising a wavelength separation unit for separating the light generated from the semiconductor device by the test pattern according to its wavelength.

8. An analysis method comprising: using a tester to input a test pattern to a semiconductor device and outputting a trigger signal indicating the start timing of the test pattern and a test result signal indicating Pass or Fail, which is the test result of the semiconductor device based on the test pattern; using a superconducting single-photon detector to detect light generated from a field of view set on the semiconductor device in response to the input of the test pattern and outputting a detection signal; using a processing unit to process the trigger signal and the detection signal to measure the elapsed time from the start timing to the timing of the detection signal and generate light emission timing information indicating the elapsed time; using the processing unit to classify the light emission timing information into either Pass or Fail based on the test result signal output from the tester; and using the processing unit to aggregate the frequency of the elapsed time indicated by the multiple light emission timing information when the tester repeatedly inputs the test pattern, for each Pass and Fail.

9. The analysis method according to claim 8, wherein the processing unit is used to subtract the frequency of the elapsed time classified as Pass from the frequency of the elapsed time classified as Fail for each elapsed time.

10. The analysis method according to claim 8 or 9, wherein a scanning unit is used to scan the field of view two-dimensionally on the semiconductor device.

11. The analysis method according to claim 10, wherein the processing unit is used to create a light emission image of a Pass at a desired elapsed time based on positional information relating to the field of view and the frequency of elapsed time in Pass, and to create a light emission image of a Fail at a desired elapsed time based on positional information relating to the field of view and the frequency of elapsed time in Fail.

12. The analysis method according to claim 11, wherein the processing unit is used to create a difference image representing the difference between the light emission image of Pass and the light emission image of Fail.

13. The analysis method according to any one of claims 8 to 12, wherein a polarization separation unit is used to separate the light generated from the semiconductor device by the test pattern according to its polarization.

14. The analysis method according to any one of claims 8 to 12, wherein a wavelength separation unit is used to separate the light generated from the semiconductor device by the test pattern according to its wavelength.

Images