Beyond silicon based devices: Is GaAs the best alternative?
#Tech #Semiconductors #GaAs #Pump-probe spectroscopy #Ultrafast phenomenon
By Clement
Semiconductors, the powerful yet fascinating materials that lies behind our day-to-day electronic devices has been the power horse of info-technology since the late 1950s.
For a long time, prominent semiconductor foundries such as TSMC uses a kind of semiconducting element called Silicon (Si) as the foundation of transistor fabrication.
Silicon, as the base material of transistor building has some great advantages such as:
• It is the second most abundant element on the earth.
• Low cost due to well-established processing.
• Large wafers can be handled safely without damage.
The famous Silicon Valley was also coined after this wonderful element that paved the way to modern small die sized computer chips.
Yet the age of silicon chips is facing grave challenges.
Since Silicon has all these benefits, why all the fuss?
If you ever read tech related articles or news, you might hear a saying: “Moore’s Law is over”.
The observation made by Gordon Moore, co-founder of Intel, has unfortunately, due to fabrication difficulties and quantum physical complications with small transistor size, slowed down to data density doubling every 18 months and its definition gradually shifted towards performance and energy-saving wise instead of sheer density of transistors.
Possible workarounds?
Scientists and engineers worked hard to break through this bottleneck, and they came up with interesting ideas such as Intel’s 3-D transistor packaging.
Optoelectronic engineers also provided some insights: why not use light as the medium of signal transfer instead of electricity? Since photons travel much faster than electrons, photo-signals can also contain higher density of information.
This is where silicon falls short of, since silicon does not easily generate photons. In other words, silicon is not an efficient LED light generator. Therefore, engineers chose a better candidate for this task: GaAs.
GaAs? What can it do?
GaAs, with the full name Gallium Arsenide, is a compound of the elements gallium and arsenic. It is a III-V semiconductor often used in IC chips, microwave IC, infrared LEDs, solar cells and much more. GaAs semiconductor devices are relatively insensitive to overheating and work at high radio frequencies. These properties are the good reasons to use them in mobile phones and satellite communication systems.
Without diving into some crazy physical and chemical characteristics of GaAs, I want to first talk about how these materials emit and absorb light, commonly known as the photoelectric effect.
Photoelectric effect is the generation of free electrons or holes when light is shined on the material. Electrons emitted in this manner can be called photoelectrons. The light particles or photons have a characteristic energy which is proportional to the frequency of the light. To avoid confusion, I will call free electrons or holes charged carriers in the following context.
The photoelectric effect is also a two-way process. If we give bounded carriers photon energy, they can hop to a higher energy state; oppositely, when charged carriers fall back to their original states, they can produce photons or light.
An interesting fact is that the famous physicist Albert Einstein received the Nobel Prize for his discovery of the law of the photoelectric effect instead of his notable work, the theories of relativity.
In our case of concern, the advantage of GaAs is that it has a direct band gap, which means that it can be used to absorb and emit light efficiently. Silicon, on the other hand, has an indirect band gap, so it requires other physical mechanisms to create photons.
What’s more?
For semiconductors, when electrons absorb abundant photon energy, it will transit to a higher energy state or the conduction band. Meanwhile if we apply a voltage bias at the sides of the semiconductor material, the charged carriers will start flowing towards the electrodes, hence forming an electrical current.
In reality, the electrons tend not to stay at the high energy conduction band for a long time. Just like most physical systems in which particles prefer to stay at a steady state. Therefore, electrons will drop back to its original state, the valance band rather quickly.
This drop back action or more scientifically speaking, carrier recombination, will take place within picoseconds up to microseconds depending on the properties of the material and other environmental factors. Once the electrons fall into the valance band, they can no longer be driven by the voltage bias, hence stopping the current flow.
By using scientific “timers” to measure the average time it takes for the electrons to recombine at the original states, we will obtain an important physical property to characterize a semiconductor material: The carrier lifetime.
Carrier lifetime is extremely sensitive to the smallest amounts of impurities or defects in the material and hence it is an ideal parameter for the characterization of material quality and process control. Throughout the twentieth century, scientists have thought of many ideas to measure the carrier lifetime of materials once the theory has matured.
In the late 1940s, physicists Haynes and Shockley conducted an experiment to measure the carrier lifetime of silicon for the first time. This experiment demonstrated that minority carriers in a semiconductor could result in a current that could be measured externally. Their results explicitly showed its capability of obtaining several crucial physical parameters: carrier lifetime, mobility and diffusion coefficient.
The ultimate switch — short carrier lifetime
In solar cells, we would wish to use a material that has a long carrier lifetime, since it gives a higher probability for carriers to form a current flow. This is also the case for electronic transistors such as BJT (Bipolar junction transistors) to provide higher current gain.
Considering optoelectronic devices, however, we will need an opposite scenario. Devices generating and receiving high frequency electro-magnetic waves such as photo-conductive antennas (PCAs) will require ultrafast current to create broadband signals. Therefore, short carrier lifetime acts like a super rapid switch to generate extremely narrow current pulses.
For semiconductors like low-temperature grown GaAs or regular GaAs, their carrier lifetime consists only of hundreds of femtoseconds up to a few nanoseconds. Which is why low-temperature GaAs works as a popular choice to fabricate PCA devices.
Pump-probe technique — Using laser to “see” photoelectrons
In order to successfully identify the short carrier lifetime of semiconductor materials, scientists developed a unique solution: The pump-probe spectroscopy.
Spectroscopy, the study of the interaction between matter and electromagnetic radiation, is a fundamental exploratory tool in the fields of science. Different spectroscopic approaches may allow the composition, physical structure and electronic structure of matter to be investigated from atomic scale to over astronomical distances.
Like I mentioned earlier, once we shine photons that provides abundant energy for electrons to hop to a high energy conduction band, the region in contact with the incident light will gain a large population of unbounded electrons. These free electrons will alter the properties of the material, such as changing the reflectivity and refractive index.
Pump-probe spectroscopy particularly is the technique of using two ultrafast laser pulses, one strong pump to excite the carriers of the material and a weaker, delayed probe to check the difference of reflectivity. This is due to photo-generated free electrons will block the passage of the probe laser rays, hence lower its transmission rate and increase the reflection.
To get the full picture of photo-generated carriers, we sample the reflection difference by changing the delay time of the pump laser and probe laser. With various delay time, the reflected probe laser at each instant could be captured and combined to map out the complete dynamics of the charged carriers.
Interestingly, the pump-probe system is not limited to merely measure picosecond or nanosecond lifetimes. Theoretically, if an experimentalist has the equipment to create long delays for the probe laser, microsecond and millisecond dynamics of materials like silicon could also be mapped out successfully.
Using the method above, ultrafast lifetime properties of excited carriers in semiconductors can be visualized with good precision, providing us a solid evidence of material quality and characteristics.
For electronic engineers, this will show whether their semiconductor samples can achieve desired performance or not.
III-V semiconductor based photonic computers
The theoretical advantages of III-V semiconductor based optical computing are numerous, including greater energy efficiency, potentially higher computing speeds, and greater information storage.
Luckily enough, high speed information transmission using light is already available with optical fibers, but such benefits can be realized on a more localized scale with optical computers.
While it is possible to simply combine optical fibers and traditional electronic computing components to get the best of both worlds. Unfortunately, there are still many difficulties in such an approach. For example, approximately 30% energy loss in converting an optical signal to electronic signal, and back again. In order to perform optical computations, a manner of performing optical logic must also be developed to create equivalents of electronic transistors.
Outlook
Is GaAs going to replace silicon? The quick answer is no, at least not in the near future.
Practically speaking, the argument in any large-scale manufacture of semiconductors is an economic one. GaAs chips are much more expensive, while silicon is cheap. If your device could work well enough with silicon, there is no need to swap out for GaAs yet.
Another decisive factor is that even though GaAs performs better for optical and radio frequency chips, it does not fit well with current CMOS technology, which is commonly used in consumer electronics.
Optical chips are still some way behind electronic chips, but a new computer revolution is steadily unfolding. With breakthroughs conceived by the academia and industry every year, we might actually live in a world where cellphones and computers powered by light in the near future.
That is when III-V semiconductors like GaAs will shine its glory.
References
https://physics.info/photoelectric/
http://www.iue.tuwien.ac.at/phd/entner/node11.html
http://www.iue.tuwien.ac.at/phd/park/node32.html
http://meroli.web.cern.ch/Lecturelifetime.html
https://en.wikipedia.org/wiki/Pump-probemicroscopy
https://www.findlight.net/blog/2019/02/01/optical-computing/
Author/Clement
Half physics, half EE senior student at NTHU. Cherish every moment in life, work hard for a better tomorrow.
