Avalanche photodiode sensors, while not an entirely new technology, are nearing a point in the development cycle where advancement hits breakneck speeds. APD tech could be where digital cameras were in the early 2000s: on the verge of dramatically increasing resolution while simultaneously falling in price, putting previously upper crust tech in the everyday consumer’s hands in a matter of years. The result would be accessible, extremely precise 3d cloud mapping capabilities applicable to industries as old as construction and new as autonomous driving - along with everything in between.
The technology is characterized by its ability to exploit the electron avalanche process when operated in certain modes. This results in unparalleled granularity of output and rapid acquisition; all within a solid-state device. Let’s look at how they work and why APDs could see a major uptick in the next five years.
Avalanche photodiodes (from here on referred to as APDs) are a sensitive type of semiconductor capable of converting miniscule optical inputs into electricity. As one would expect, this level of sensitivity to light doesn’t come without some complication. There are several clever exploits APDs employ to extract the desired result from nature. To understand APDs we must first look at the descriptively-named electron avalanche.
Discovered via the work of Irish physicist John Townsend at the turn of the 20th century, the electron avalanche is the result of impact ionization (i.e. the mobilizing of once-bound electrons) by kinetic energy. When this occurs within the realm of a sufficiently strong electrical field the resulting increase in kinetic energy can escalate these events into avalanche breakdowns.
The avalanche breakdown process can be imagined as dominoes arranged in a triangular formation. If two dominoes are placed side-by-side in front of one, they can both be toppled simultaneously by tipping the first, then those two simultaneously toppling four, and so on. While tipping a straight line of dominoes is a clear expression of Newtonian action-reaction, one domino tipping two at once in an expanding triangle (avalanche) has the apparent effect of doubling kinetic energy each time the force’s path branches. While we know the basic principles of Newtonian law do not permit this, an environmental force can be recruited to achieve a similar effect.
In this case, the original application of energy (representing the photon input) consists of your tipping of the first domino. The moment your finger ceases to exert force on the piece, gravity is already taking over and facilitating your tyrannical scheme to topple all the pieces. While a small amount of kinetic energy is lost every time, the dominos collide (think air resistance and absorption of energy during impacts) this loss is far outweighed by the force gravity successfully exerts on the pieces each time they lose their balance. The net outcome is as if you had multiplied the original impact. You just needed an environmental factor to help it along. In the absence of gravity, the kinetic energy transfer would have been increasingly dampened with each strike and thus failed to produce a true “avalanche” effect.
In the case of electron avalanche breakdowns, the electrical field works the way gravity did for the dominos. The presence of the field increases the electron’s kinetic energy, allowing the potential for impact ionization to escalate in the form of an electron avalanche. Electrons need a strong electrical field to foster an ability to create a greater impact on the world around them. If the field is strong enough, eventually an electron may even go on to conduct a TED Talk, impact the world with its big ideas, acquire funding for a startup poised to impact an industry, mishandle funds, misrepresent profitability to shareholders and ultimately revert to an ignominious life as a bowtie-clad talking head who guests on CNBC.
But not every electron in the realm of a strong electrical field has such a complex career. In fact, not every photon that contacts the sensor results in an electron being converted. Depending on the mode the sensor is operated in, the quantum efficiency, which identifies the percentage of input photons that produce a converted electron, can be 30%. While this number may not sound impressive at first blush, consider that the result is an average of just over three photons to result in a signal. Contrast that with photographic film which has a quantum efficiency well below 10%, meaning that a given photon is converted to an electronic charge about as often as a family reunion is a pleasant experience.
This amplified signal produced by the avalanche breakdowns creates an interesting situation: It allows otherwise insular media to be conductive within the scope of these breakdowns, resulting in an inherently higher gain for the desired signal that stands out from unwanted and omnipresent quantum dark noise.
Picture an audio engineer who is planning to record a classical pianist’s concert hall performance with highly sensitive microphones. Being experienced, the engineer knows that there will always be someone with tuberculosis in the front row of any classical concert. So, to minimize unwanted noise, he gets the mics as close as possible to the piano. The sound of the coughing still exists, but the more signal he can get from the piano the less problematic the background noise is.
Quantum dark noise works in a similar way. The dark noise comes from unwanted free electrons roaming around generated from thermal energy as opposed to photons. But increasing the desired signal via avalanche breakdowns suppresses the noise’s effect.
While the APD does a great job of mitigating dark noise, there remains the issue of optical cross-talk by means of internal light reflections. There are potential physical and algorithmic means for preventing this from contaminating the final output, but as detector arrays increase in size to have higher resolution, effective optical isolation becomes increasingly difficult and new methods will need to be adopted to make these larger sensors practical. This represents one of the primary challenges in cost-effective production of APDs.
But the challenge will likely be worthwhile. To be useful in real-time applications (augmented reality, for example) range finding systems have to be capable of constructing data rapidly, which is precisely where APDs can excel. With the capability for higher point acquisition speeds at longer ranges than conventional methods, the technology could quickly find uses in previously un-breached fields.
The next five years could see a significant shift in APD development. As resolution and affordability increase in unison, don’t be surprised if the sensors and their associated tech begin to crop up in unexpected places much the same way digital cameras found themselves a staple feature for cell phones. APDs might just bury the competition.
About the author
Ryan LaDue is an author who specializes in topics relating to construction technology and data science. He is based in Nashville, TN.