Making Sense of BSI Sensors
An image sensor can be called the heart of any digital camera. Being technically complex, sensors are never fully understood by most people. Here we try to demystify the world of sensors, from CCDs to Back-side Illuminated (BSI) CMOS sensors without elaborating deeply on the technical aspects. To explain the different types of sensors, it is important to understand the functioning of sensors. The image sensor is the last point of contact for light after passing through many layers of glass (lens elements) and the colour filter array. Each photo receptor (photo diode) on the sensor converts the light striking it into an electrical charge proportional to the light intensity at that spot, much like a solar cell on a solar panel. But unlike in a solar cell, the electric charge thus generated is stored in the diodes till it is read by another circuitry.
Charge-Coupled Device (CCD) Sensor
In the case of a CCD sensor, the electrical charge is transferred across the sensor towards one corner and an amplifier converts the charge into voltage. This process is repeated till the entire array has been read out in the form of a sequence of voltages. An Analogue-to-Digital Converter (known as ADC, A to D or A/D) then converts the voltage into digital information.
Complementary Metal Oxide Semiconductor (CMOS) Sensor
A CMOS sensor is an active-pixel sensor imaging device. This means that each pixel on a CMOS sensor is connected to an active amplifier, which converts accumulated electrical charge from each pixel into a voltage before it is transferred to the ADC. Both the CCD and CMOS have their share of advantages as well as limitations. There is no clear
advantage of using either of the technologies. CCD sensors produce high-quality, low noise images compared to CMOS sensors. CCD sensors consume approximately 100 times more power than CMOS sensors. CMOS sensors are less expensive to manufacture. CCD sensors also suffer from vertical smear and blooming from bright light sources. CMOS sensors process information from a row at a time, while CCD sensors capture the entire image at once. This can cause a rolling shutter effect, which distorts the image in videos. Hence manufacturers have been constantly at work to improve the efficiency of CMOS sensors to a level similar to CCDs. As mentioned before, CMOS sensors require to have several auxiliary components (such as transistors) next to each pixel. This circuit layer is positioned on the front surface (this can be visualised as photo receptors peeping out of gaps left by the circuit layer), severely affecting the efficiency (or light gathering capacity) of the pixel layer (photo receptors) below. To overcome this, manufacturers of CMOS sensors use micro lenses to direct more light onto each pixel so as to improve the efficiency of the sensors. But this also has limitations since the circuit layer (also referred to as ‘illumination’) reflects part of the incident light, reducing the quantity of light falling on the pixel layer. The megapixel war necessitated every manufacturer to increase the pixel count on the sensor by cramming more photosites. But there is a practical limit to the number of photosites that can be fabricated on a sensor, and the circuit layer complicates the issue by using up considerable amount of space. The solution was to move the circuit layer to the back of the sensor unit.
Back-side Illuminated CMOS
Manufacturers have now found a way to deposit (or fabricate) the circuit layer at the back of the pixel layer. Such a sensor is known as Back-side Illuminated (BSI) sensor. Since the pixel layer is in front of the circuit layer, a BSI sensor receives more light per photo diode and hence the efficiency is higher. Also, the frontside illuminated (FSI) sensor carries less number of photo receptors than their similarly-sized back-illuminated counterparts since the circuit layer blocks part of the sensor area. Another advantage of the BSI sensor is that a smaller sensor can pack more photo receptors because more surface area is freed due to the shifting of circuit layer to the back side, thus increasing the pixel resolution (megapixels) of small sensors without affecting the low-light performance. This is quite an advantage at this time of small imaging devices (such as Smartphones) demanding higher ‘megapixels’. Since photo receptors on BSI sensors receive more light than their FSI counterparts, their low-light performance is better, producing more detail and generating lower noise. Sony was the first to introduce BSI CMOS sensors and they call it the Exmor R CMOS sensor.
Stacked CMOS
Fabricating the circuit layer at the back of the same substrate layer (the base material on which different components are deposited or integrated) posed some serious challenges to manufacturers. For example, trying to improve the light gathering capacity of the pixel layer damaged the silicon substrate layers, and hence they needed to be heat-treated to heal the damages. But this heat treatment reduced the efficiency of the circuit layer. Again, the manufacturing process made the substrate very thin, requiring a supporting substrate to be used to strengthen it for transportation. Hence Sony decided to integrate the circuit layer on the supporting substrate, which is also a silicon substrate identical to the one used for the pixel layer and stack the two layers together, one over the other. This stacking of layers solved all the problems associated with the manufacturing process, and reduced the size of sensor considerably due to the reduction in auxiliary circuits used to improve the efficiency of BSI CMOS. Sony named the new sensor Exmor RS, and it is now widely used in mobile devices. Backside illuminated sensors have been in existence for quite a long time, especially in high-end applications such as astro photography and surveillance equipment. Then why weren’t they used in commercial photographic equipment? The reason was that these sensors were very expensive owing to the complex process by which the silicon wafers were produced. But now the demand for BSI sensors have gone up many fold and hence the price has begun to fall considerably. The new technologies have culminated in better light gathering capability, higher dynamic range, and compact size of image sensors. Though semiconductor technology is not as simple as it is explained here, diving deeper into the fabrication process might not be of interest to many of our readers and hence could be beyond the scope of this feature.