Quantum efficiency spectrum and defect analysis (continued)
- This article explains the quantum efficiency spectrum and four common CMOS image sensor process defects.
- Wafer-level testing equipment can provide more detailed defect information than traditional optical inspection tools, which helps characterize CMOS image sensor performance. The quantum efficiency spectrum is a key parameter that reflects the sensor's sensitivity across wavelengths and thus affects image quality.
C. Color filter quality inspection
1. What is the color filter in an image sensor?
The color filter in an image sensor is an optical filter applied to CMOS image sensors to shape each pixel's spectral response so the sensor can detect and separate different colors and convert them into a digital signal. Color filters typically include red (R), green (G), and blue (B) elements. Different color filter array (CFA) arrangements exist; the most common CFA is the Bayer pattern, in which each 2x2 unit contains one B, one R, and two G filters.
The color filter plays a critical role in color reproduction. Its quality directly affects color accuracy. Precise spectral analysis and quality inspection are required to ensure the color filter meets design requirements. Transmission spectra assess optical performance of the filters, while the quantum efficiency spectrum evaluates matching between the color filter and the photodiode. Strict quality control helps ensure consistent image output.
Figure 1. How color filters are combined in a pixel. A pixel unit consists primarily of a micro lens, the CFA, and the photodiode.
The main function of the color filter is to decompose incident white light into spectral components and selectively transmit or block certain wavelengths to achieve wavelength-selective sensitivity.
2. Why is color filter inspection important?
Each pixel contains a color filter that selectively senses R, G, or B light to capture color images. Poor color filter performance affects pixel sensitivity and spectral response, degrading image quality and accuracy. Color filter defects or nonuniformity can cause color shifts, distortion, and uneven color rendering. Rigorous inspection ensures filter performance and stability meet design requirements, improving sensor yield and reliability.
3. How to use the quantum efficiency spectrum to assess color filter quality?
Color filters are commonly made from organic dyes or pigments that selectively absorb specific wavelengths. These dyes are typically deposited by coating onto glass or silicon substrates to form the color filters.
The quantum efficiency spectrum measures sensor sensitivity across wavelengths and can be used to assess color filter performance. A properly designed color filter should separate wavelengths with minimal optical crosstalk. If the quantum efficiency at a given wavelength is lower than expected, it may indicate color filter performance issues. Analysis of the quantum efficiency spectrum allows verification of filter performance against design targets and guides adjustments and optimization.
Figure 2. Wafer-level quantum efficiency spectra for three green filter materials (Green_1, Green_2, Green_3). The spectra are used to evaluate sensitivity and optical crosstalk effects.
Comparing these materials shows:
- The primary green peak shifts to 550 nm;
- Optical crosstalk from green into the blue channel is significantly reduced;
- Optical crosstalk from green into the red channel is significantly increased.
Quantum efficiency analysis verifies whether the filter material meets design requirements and guides adjustments to ensure image fidelity and sensor reliability.
D. Silicon wafer thickness control
1. What is silicon wafer thickness control?
In backside-illuminated (BSI) CMOS image sensor manufacturing, a thinning process reduces the wafer thickness. The thinned wafer thickness directly affects sensor sensitivity, so precise control of wafer thickness is important for sensor performance and quality.
Silicon wafer thickness control processes are used to achieve the target thickness after thinning, ensuring the sensor meets design specifications and improving production yield.
Figure 3. BSI process flow. An important step is the thinning operation, which reduces wafer thickness to a specified value.
2. Why is quantum efficiency testing important for wafer thickness control?
Wafer thickness directly affects sensor sensitivity, and this impact can be observed using the quantum efficiency spectrum to ensure the thinned wafer achieves optimal photoelectric conversion efficiency. Testing at wavelengths of 450 nm, 530 nm, and 600 nm evaluates the blue, green, and red channels respectively. Experimental results show how quantum efficiency varies with different thinning depths. Deviations in thinning thickness directly affect sensor sensitivity and quantum efficiency, so quantum efficiency testing is critical for monitoring wafer thickness control processes to ensure consistent sensor quality.
Figure 3 shows quantum efficiency variation in the blue, green, and red channels at different thinning thicknesses. The blue channel was measured at 450 nm: when the thinned thickness was 0.3 um thicker than the standard, quantum efficiency dropped from 52% to 49%; when 0.3 um thinner than the standard, the blue channel quantum efficiency remained close to 52%. The red channel, measured at 600 nm, behaved oppositely: when the thinned thickness was 0.3 um thinner than the standard, red channel quantum efficiency dropped significantly from 44% to 41%; at +0.3 um it showed no significant change. The green channel, measured at 530 nm, showed no significant change within the STD THK ± 0.3 um range.
The results indicate that silicon wafer thickness significantly affects quantum efficiency, with different channels affected to different extents. Precise thickness control during manufacturing is therefore essential to ensure sensor performance and product quality.
