You’ve got your subject in the viewfinder and the frame is composed. As you press the shutter button, a suprisingly complex process ensues. First, light energy which spent its last eight minutes traveling from the Sun hits its Earthly target and reflects into your lens. Then photons channel their way through the glass and smack the surface of your sensor. You might think that the journey of the photon would be over at this point, but this is only where the fun begins.
The journey of a photon from the surface of the sensor to the actual light sensing photosite is a dangerous adventure. The path is so treacherous that less than 50% of the photons which start the journey survive. When there is plenty of light, you probably don’t even notice all the casualties. But in dark conditions, you need every photon you can get. Sensor designs which provide the easiest obstacle free light path are rewarded with better low light performance. In this article, we’ll describe five ways sensor makers encourage better photon survival and as a consequence, obtain much better higher ISO performance.
1. The Gapless Microlens
One major light-loss problem in the image sensor is a suprisingly mundane one. The problem is how to wire all the pixels together without the wiring getting in the way of the light path. As shown in the diagram below, the actual light sensing diodes are nestled innocently at the bottom of a big layer cake. All the overlying layers are needed in order to route electrical signals throughout the chip and on to the outside world. The wiring must be creatively routed so that it doesn’t block the light path to the photosite below. The more pixels that are crammed onto a sensor, however, the harder this is to do. On a modern sensor, some of the light will invariably be blocked by the wiring.
Enter the microlens. The micro lens concept was invented to combat the light blocking problem by simply steering the light around the obstacles below. Almost all modern cameras at this point incorporate some kind of microlens atop the sensor. As technologists continue to push for better low light performance, they have discovered better ways to design the microlens for maximum light collection.
One idea is the gapless microlens. In days of yore, micro lenses were fabricated by patterning small liquidous squares atop each pixel. The squares were then heated such that the liquid balled up into tiny lens shaped hemispheres. The drawback of this technique was that there were limits on how closely the lenses could be formed. If the lenses were patterned too closely together, the liquidous squares would flow together and lose their shapes. The result was that there were gaps between the microlenses in which vital light energy would fall undetected. In recent years however, the patent offices have been flooded with more creative concepts for fabricating microlenses which are “gapless.”
Canon first introduced the gapless microlens with its 50D camera and then successfully used them on the higher pixel 7D camera. The new flagship 1D X camera is Canon’s first incorporation of gapless microlenses on a full frame sensor. Nikon uses a similar gapless microlens concept on their D3 series cameras, as well as the D700.
2. Better Light Path Aspect Ratio
One advantage of the CMOS sensor is that it benefits from the technology advances realized by other high volume CMOS processes. Late in the 1990′s Intel, IBM and others pioneered the use of copper wiring within their microchips to provide better performance on their ever more complex microprocessors. Aluminum had previously been used due to the relative ease with which it could be patterned, but copper had the advantage of providing better current carrying capacity in a much smaller space. For image sensors, the move to copper wiring has an added benefit – reduced stack height. As shown in the diagram below, a copper system has a shorter stack height which allows light collection from a much broader array of angles.
Another take on the light steering strategy is to excavate columnar cavities through which light can be directed. The sides of these cavities are then lined with reflective materials such that light is bounced downward with minimal loss. This approach is similar to a fiber optic cable which guides light through a series of total internal reflections.
4. Backside Illumination
As mentioned in a previous article, some vendors are choosing to avoid the light blockage problem altogether by flipping the sensor over and allowing light to reach photodiodes from the backside. This is a delicate and costly operation which has so far been limited to smaller size sensors. It also creates new problems of crosstalk between pixels. But backside illumination may ultimately be the path towards ultra high efficiency sensors.
5. Photodiode Engineering
If light manages to make its way to the bottom of the layer cake, the last obstacle lies in the efficiency with which the photodiodes convert light into electrical signals. Progress in photodiode engineering has been slow but steady over the last twenty years. As described in this article, photodiodes are formed by subtly manipulating atomic layers of silicon in just the right ways to achieve optimal capture rates. As the years have gone by, more and more complex schemes have been devised for capturing light with better efficiency. We can expect a continued push for incremental improvements in this area.
The Bottom Line
The last ten years of image sensor marketing can be characterized as a race for megapixels – mostly because consumers have been trained into using pixel count as a primary comparison metric. Recently, though, we’ve seen more of a focus on low light performance as the true measure of a camera’s greatness. While ISO performance is not as easy to measure as the number of pixels, it should be an important metric to consider. With the technologies camera makers have in their back pocket, consumers have the right to expect greatness. So go ahead – crank up the ISO on that next new camera. You might just be surprised at what you see.
Get More Camera Technica Greatness: