In today’s tech world, CMOS Image Sensors (CIS) have permeated every corner of our lives, from smartphones to self-driving cars. CIS’s influence is increasingly significant. However, to ensure the performance and quality of CIS, we need to conduct a series of wafer-level tests. These test methods include functional testing, performance testing, and reliability testing, each playing a pivotal role in ensuring the performance and lifetime of the chips.
Functional testing verifies the basic functionality of the CIS and detects early defects. It includes tests of interface ports, registers, timing, image quality, and other key functions. Performance testing quantifies critical parameters like sensitivity, dynamic range, noise, linearity, uniformity, etc. It provides an objective measure of the sensor’s capabilities. Reliability testing evaluates the CIS’s durability under various stresses like temperature, voltage, vibration, etc. It ensures stable operation over the product lifetime.
Semiconductor testing plays an extremely important role in the four key steps of semiconductor design, manufacturing, packaging and testing. It utilizes specific test equipment to examine the device under test (DUT) and basically involves test program generation, applying test vectors to the DUT, capturing response data, and comparing to expected results to determine pass/fail status.
It can identify product defects, verify if the product meets design goals, and sort good dies from bad dies. In recent years, more and more manufacturers start to care about measuring if product performance matches the intended design specs. Therefore, test equipment is irreplaceably crucial from design validation to the entire semiconductor fabrication process.
However, according to the 10x rule of electronic system testing, if a chip’s quality or performance issues are not caught during manufacturing, it will incur 10x higher cost to discover and resolve the problems at the PCB level. This highlights the importance of testing, especially the benefits of early detection.
Since semiconductor testing is a huge industry chain system, its test classifications are extremely complex. This article will guide everyone to break down the entire semiconductor test industry chain from process flow and test purpose perspectives.
In general, semiconductor chip fabrication can be divided into three main stages – “IC design” → “wafer manufacturing” → “packaging & testing”.
Correspondingly, the tests can be categorized into three types – IC design verification, front-end wafer testing, and back-end final testing.
“IC design verification → IC design stage”
“Front-end wafer testing (in-process inspection & wafer sort) → wafer manufacturing stage”
“Back-end final testing → packaging & test stage”
Design Verification (IC Design Stage): Used during IC design phase, also known as lab testing or characterization. It verifies if the design is correct before entering mass production. Requires functional testing and physical validation, mainly using electrical measurements to validate if the device achieves the intended functionality per the design.
Front-end Testing (IC Manufacturing Stage): Used during wafer fabrication process. Employs optical inspection, e-beam metrology etc. to perform functional and performance tests, and check for defects on the wafer that may impact yield.
Back-end Testing (IC Packaging Stage): Mainly used after wafer fabrication and during IC packaging process. Uses probe stations, testers, sorters etc. to conduct electrical and functional tests, ensuring the subsequent packaged chips meet specifications. Back-end testing can be further divided into two categories:
CP Wafer Testing: Ensures wafer quality and performance meet requirements to enter packaging flow.
FT Final Testing: Validates quality of finished packaged chips.
During the process of “design verification testing”, “pre-production inspection/measurement”, and “post-production inspection”, products must go through various complex inspection and measurement procedures at each stage. If we break down these inspection and measurement methods and reclassify them by their functional goals, they can be divided into three major types: “functional testing”, “performance testing”, and “reliability testing”.
Most of the “functional testing” and “reliability testing” methods are designed to detect physical or electrical defects in the device under test. However, performance testing in post-production inspection is specifically used to evaluate the actual functional performance of the product being tested.
In the CIS field, how can we understand the performance of image sensors through functional testing? This is where the EMVA1288 standard introduced by the European Machine Vision Association (EMVA) comes into play. This standard specifies various image quality parameters, as well as corresponding measurement and representation methods, allowing us to quickly understand the performance differences between various sensors.
In the following articles, we will take a deep dive into these testing methods, and explain how to use the EMVA1288 parameter standard to evaluate sensor performance. We will also introduce some commercially available test equipment, including Enlitech Technology’s SG-A and SG-O series products. These systems are designed for wafer-level testing and can provide comprehensive and accurate test results.
By leveraging advanced equipment and standardized methods, we can thoroughly characterize critical performance attributes like quantum efficiency, temporal noise, and dynamic range. This enables objective comparisons between different sensor architectures and process technologies. With a strong understanding of performance benchmarks, designers can make informed decisions to optimize their next generation image sensors. We look forward to sharing our insights on measuring and maximizing CIS quality and reliability.
From Physical Inspection to Electrical Testing to Performance Verification
Although AOI inspection offers many benefits for CIS wafer fabrication, it also has some limitations. Since CIS wafers exhibit diverse surface properties and defect types, and are also influenced by environmental factors like illumination, different optical settings and algorithms are required for different CIS wafer types. At the same time, continuous calibration and maintenance of the AOI inspection system across the production line is needed to ensure accurate and consistent defect detection.
Automatic Optical Inspection (AOI) is an imaging-based method to detect defects on CIS image sensor wafers. The system contains an active light source that illuminates the CIS wafer surface. The reflected light is collected by a camera lens, forming an image of the CIS surface. This image can be analyzed and processed by imaging software to automatically identify defects on the CIS wafer surface.
The advantages of AOI inspection are that it can efficiently and quickly detect defects on the CIS wafer surface, including flaws, contaminants, particles, etc. The optical inspection process typically involves three steps: illumination, imaging, and analysis.
First, we need to use a proper light source to illuminate the wafer surface. The choice of light source is crucial for the accuracy and effectiveness of optical inspection. Different light sources like LED or laser can provide varying light intensity and wavelength to meet different inspection requirements. During illumination, we need to ensure uniformity and stability of the light spot to obtain a clear image.
Next, we use optical instruments to capture images of the wafer surface. Common optical instruments include microscopes, CCD cameras, and optical inspection systems. These instruments can magnify and digitize the wafer surface image for subsequent analysis and processing. The instrument selection should be based on inspection accuracy requirements, imaging speed, system reliability, etc.
Finally, we analyze the obtained images to identify potential defects. The analysis may involve techniques like image processing, image comparison, and defect annotation. With these analysis techniques, we can quickly and accurately detect defects on the wafer surface, such as scratches, pinholes, or stains. These defects can adversely impact wafer fabrication by affecting device performance or reducing product yield. Well-known vendors include KLA-Tencor, Applied Materials, Visera, Hitachi High-Tech, etc.
Although AOI offers many benefits for CIS wafer production, it also has some limitations. Since CIS wafers exhibit diverse surface properties and defect types, and are also influenced by environmental factors like illumination, different optical settings and algorithms are required for different CIS wafer types. At the same time, continuous calibration and maintenance of the AOI inspection system across the production line is needed to ensure accurate and consistent defect detection.
Electrical testing is a method to inspect the electrical performance of wafers using electrical instrumentation. The importance of this method lies in ensuring the electrical performance of wafers meets design requirements. The process uses electrical instrumentation to inspect the electrical performance of wafers.
Electrical testing can measure parameters like resistance, capacitance, inductance, current, voltage, frequency of circuit elements and interconnects on the wafer, as well as logic functions and digital signals. These parameters and signals are critical factors affecting wafer performance. Failure to meet requirements can lead to wafer failure or product defects.
The electrical testing process usually involves three steps: configuring test conditions, conducting tests, and analyzing results. First, we need to set appropriate test conditions like voltage, current, and frequency based on the wafer specifications and design requirements. These conditions affect the sensitivity and accuracy of the electrical instrumentation, as well as the stress and thermal effects on the wafer. Therefore, suitable instrumentation and probe cards must be chosen and calibrated. Next, tests are conducted using the electrical instrumentation, typically by bringing probe cards in contact with specific locations on the wafer and reading the corresponding data via the instrumentation. This process may need to be repeated multiple times to cover all areas and components requiring testing. Finally, the test results are analyzed to determine if the wafer’s electrical performance meets requirements. If any non-conformities or abnormalities are found, root causes must be identified and rectified.
Electrical testing enables comprehensive and accurate assessment of the wafer’s electrical performance, thereby improving wafer yield and quality. It also facilitates design and process optimization through feedback to enhance wafer performance and efficiency.
As the name suggests, functional testing primarily examines whether the wafer functions meet design requirements.
The functional testing process involves conducting specific functionality tests on the wafer to ensure it can operate normally in actual use. For example, for memory wafers, we perform extensive read/write tests to ensure data can be correctly saved and retrieved. For logic gate wafers, we do switch testing to confirm they execute appropriate operations under different logic states.
Environmental testing is a method to evaluate wafer performance by simulating actual use conditions. Its importance lies in ensuring wafer performance and lifetime under real-world operating environments. The environmental testing process typically involves three steps: configuring test environments, conducting tests, and analyzing results. First, we need to set up simulated environments including temperature, humidity, pressure, etc. Then, we perform tests under these simulated conditions. Finally, we analyze the test results to determine wafer performance and lifetime in actual use.
Through these testing techniques, we can assess wafer performance and quality from different perspectives to ensure final product reliability and lifetime.
Performance Testing: A Critical Aspect of CIS Testing
After Establishing Common Metrics, the True Performance is Revealed
In the CIS wafer manufacturing process, in order to enable clear communication and quantification of CIS performance between research, production, and design teams, the international organization European Machine Vision Association (EMVA), which has long focused on the machine vision field, formulated a set of standardized parameters for evaluating image sensor performance, known as the EMVA1288 standard. This standard defines several aspects for assessing image sensor performance:
- Quantum efficiency: The efficiency of an image sensor in converting photons into electrons. Higher quantum efficiency leads to higher sensor sensitivity and more light capture.
- Dark current: The current generated by an image sensor without illumination. Lower dark current results in lower noise and better image quality.
- Signal-to-noise ratio: The ratio between image signal strength and noise strength. Higher SNR indicates better image quality.
- Dynamic range: The ratio between the brightest and darkest luminance an image can depict. Wider dynamic range allows more detail to be captured.
- Resolution: The smallest object size an image sensor can resolve. Higher resolution enables more fine details to be distinguished.
- Contrast: The luminance difference between dark and bright areas in an image. Higher contrast provides greater sense of depth.
- Color reproduction: How accurately an image sensor reproduces the true colors of objects. Better color reproduction increases realism.
Based on these evaluation aspects, EMVA1288 further defines specific parameters to quantify image sensor performance, including the following measurement standards:
|No.||Performance Indicator||Parameter Meaning|
|1.||Quantum efficiency (QE)||Quantum efficiency (QE) measures the efficiency of a pixel in converting incident photons into charge carriers (electrons) at a given wavelength (nm). It is also the efficiency of a CIS in converting photons into electrons.|
|2.||Spectral response||The spectral response curve shows the conversion efficiency of the photosensitive element for photons of different wavelengths, indicating the sensitivity response of the photosensitive element to light of various wavelengths.|
|3.||System Gain||Gain of the imaging system on the charge or voltage signal generated by the photosensitive element|
|4.||Sensitivity||The absolute sensitivity threshold is the minimum illumination intensity required for the CIS to start generating a visible image. Lower absolute sensitivity threshold indicates higher CIS sensitivity and ability to capture weaker light images. Minimum ability of an imaging system to generate image signal for a specific wavelength and incident light intensity, in units of lux-seconds.|
|5.||Dynamic range||The ratio range of incident light intensities that an imaging system can effectively distinguish without saturation or noise, wider dynamic range allows more details to be captured in the image.|
|6||Dark Current/Noise||Current/noise generated by the photosensitive element itself due to temperature and structure in total darkness. Lower dark noise indicates better CIS image quality.|
|7.||SNR||Ratio of image signal charge to noise charge generated by the photosensitive element, representing image quality and discernibility. Higher SNR indicates better image quality.|
|8||Saturation Capacity||Maximum incident light intensity that the photosensitive element can accept before generating saturated output.|
|9||Linearity Error, LE||Percentage deviation between actual output signal and ideal linear output signal.|
|10||DCNU||Non-uniformity caused by variance in dark current at different locations of the same photosensitive element. Lower dark signal non-uniformity indicates better CIS image quality.|
|11||PRNU||Non-uniformity caused by variance in current response of the same photosensitive element to the same incident light at different locations. Lower photon response non-uniformity indicates better CIS image quality.|
|12||CRA||The maximum angle of light focused onto a pixel from the sensor side of the lens is defined as a parameter called chief ray angle (CRA).|
The availability of these performance parameters and measurement methods helps design, production, and customer teams quantify and evaluate image sensor performance in concrete ways. It also serves as a benchmark for comparing image sensor capabilities between different manufacturers.
Through 20 years of effort, the EMVA1288 standard has gained widespread global recognition and become an important benchmark for measuring image sensor performance. It provides the industry with a common language and framework, holding significant reference value in market competition and product selection.
Leading global companies and academic institutions in CIS like Sony, Canon, Panasonic, Samsung, Infineon, Texas Instruments, Allied Vision Technologies GmbH, Enlitech, Basler AG, Cognex Corporation, Fraunhofer-Gesellschaft are EMVA members. The association’s diverse membership encompasses industry, government and academia across the entire CIS ecosystem including upstream, midstream and downstream players as well as related equipment producers.
Since the introduction of the EMVA1288 standard by EMVA, there has been global consensus and a common language for image sensor design, manufacturing, and testing. Especially EMVA1288 Edition 4 released in 2021 has become almost like the inspection bible for many CIS designers, manufacturers, and foundries. In numerous procurement contracts, related parameter values are defined as acceptance criteria. Therefore, efficiently and fully measuring key EMVA1288 parameters like quantum efficiency, dark current, and SNR has become a major pain point for companies.
While domain expertise is plentiful, truly cross-disciplinary integration capabilities are rare
Previously, due to the lack of integrated equipment and varying standards between companies, no vendors sold complete integrated measurement systems. At best, they offered modular components. So companies needing testing had to buy parts and build their own custom systems. Although components were readily available, system integration required high expertise across optics, optoelectronics, mechanics, and informatics. Developing these systems needed multiple engineers across disciplines and incurred immense time and cost. Moreover, self-built systems often suffered from poor repeatability, accuracy, and reliability due to technical challenges across the domains. Extensive recalibration and tweaking was needed with no guarantee of correct results meeting international standards or customer scrutiny.
Concurrently, few companies possessed expertise spanning optics, optoelectronics, mechanical engineering, and software engineering, along with the ability to overcome cross-domain complexities at an economically viable production scale. Most vendors focused on selling components within their domain, leaving end-user needs for complete solutions unaddressed.
Why build your own makeshift test equipment when professional integrated tools are available?
Compared to traditional CIS optical inspection, Enlitech’s SG series optical inspection solutions can:
a) Provide more comprehensive defect inspection information beyond conventional non-destructive optical techniques.
b) Help users gain a more complete understanding of CIS image sensor performance.
Enlitech is one of the early founding members of the European Machine Vision Association (EMVA) and has long focused on designing and manufacturing wafer-level CIS inspection and optoelectronic metrology equipment.(https://www.emva.org/members/enli-technology-co-ltd/)
Enlitech is a leading company in optoelectronic conversion metrology and one of the few possessing expertise across optics, optoelectronics, mechanical engineering, software design, probe card integration, etc. With decades invested in non-destructive and optical inspection techniques for CIS and CCM, Enlitech provides complete test solutions covering the entire industry chain from CIS wafers, CIS dies, CIS modules to CIS cameras.
Recognizing the difficulties faced by CIS companies in building their own systems to measure EMVA1288 parameters, Enlitech recently integrated its expertise across optoelectronics, mechanical engineering, and software to launch the SG series – the world’s first integrated commercial testers purpose-built for EMVA1288 metrology.
Enlitech’s SG-O and SG-A solutions can measure parameters including:
This series comprises three models – SG-O, SG-A, and SPD2200. Among them, the SG-O integrating probe card capabilities for wafer-level testing has garnered huge interest from major foundries, with orders already placed by the top three smartphone manufacturers.
Additionally, the SPD2200 uniquely designed for SPAD performance metrology was prominently featured in a recent article by Taiwan’s Economic Daily News and has gained adoption by a top 5 international foundry.
Enlitech’s SG-O integrated commercial test solution not only enables consolidated metrology of the various EMVA1288 parameters critical for evaluating CIS imaging performance, but can also integrate probe cards for direct data measurement at the wafer level, and be deployed within existing industry or lab settings.
The “programmable auto prober” offers tremendous flexibility to integrate hardware control panels and automate wafer loading, enabling smart wafer mapping capabilities for wafer or die-level measurements and extremely precise, reliable DC/CV and RF measurements.
To accommodate different metrology environments, Enlitech’s SG-O integrated commercial test solution employs a modular cartridge design with advanced CDA thermal control technology for high ramp rates and temperature stability. This enables operation across an ultra-wide -60°C to +200°C range along with ultra-low noise, making it highly suited for precise wafer-level testing of CIS, ALS, and optical sensors.
The SG-O’s illumination source provides an ultra-broad spectral range spanning UV to SWIR (400-1700 nm). With >98% uniformity and >98% stability over 10+ hours across a 50mm x 50mm area, and 140 dB total light intensity dynamic range, it meets the most demanding test requirements for diverse sensor types.
In our exploration of semiconductor technologies in depth and breadth, the importance of CIS wafer-level testing is self-evident. From optical inspection, electrical testing, functional testing to environmental testing, each test method plays a critical role in ensuring wafer quality and performance. These test methods not only help us determine the physical and electrical characteristics of the wafer, but that alone is far from enough. In today’s pursuit of ever greater capabilities, evaluating product performance under various environmental conditions will be the key to products breaking out of the red ocean into the blue ocean.
EMVA1288 is an international standard for measuring and comparing the performance of digital image sensors and cameras. It contains a range of tests for the optical performance, noise, dynamic range, color rendition, spatial frequency response, etc. of sensors.
Introducing standards like EMVA1288 at the wafer level can provide several benefits:
Although introducing wafer-level EMVA1288 testing increases initial inspection costs, it is well worth the investment long term. Catching problems early significantly reduces development risks and time-to-market, while increasing customer confidence. The benefits and competitive advantages are tremendous.
Under this concept, Enlitech’s SG series metrology tools showcase high integration honoring a “Let us handle the measurements so you can focus on developing products” design philosophy. The SG-O, SG-A, and SPD2200 are the only instruments purpose-built for EMVA1288 metrology. Beyond comprehensive wafer-level testing, they measure EMVA1288 parameters to provide a complete, accurate and reliable solution. Careful consideration for user needs and convenience also allows rapid deployment in existing production lines or labs, delivering stable, efficient, precise testing. By choosing integrated high-precision metrology tools, you can finally break free from the nightmare of building your own systems and accelerate breakthrough developments.
#SG-A #SG-O #SPD2200 #CIS #CMOS #LiDar #Sensor