TE Connectivity

As a Segment CTO and VP, Industrial Solutions Segment at TE Connectivity, Erin Byrne guides a diverse, global team of deep technical experts in multiple sensing modalities toward targeted growth applications that require differentiated performance, reliability, and scale.

What are some of the most significant challenges that companies face in leveraging Optoelectronics?

I believe that scaling up is a challenge for industries with manufacturing roots. Optics operates in three dimensions and at smaller scales than a semiconductor-type process. Therefore, dimensional control is necessary to construct the product. This involves some sort of alignment as it is so precise; it either needs to be done automatically or takes a little bit longer. This is a manufacturing cycle time restriction that prevents the production of millions of these items and getting lower cost measures. Therefore, in optoelectronics, you are frequently fighting to create a quicker and smaller ideal electronic version vs. something you might be able to do electrically. But that probably takes up more space and may not be as quick. Optics hence has the advantage of being resistant to many EMC fields. However, the challenges we face include integrating the fiber and handling it inside a product to send the data around rather than using a PCB trace.

The second challenge is removing heat. During the process of changing electricity to light and light to electricity, conversions are inefficient enough not to generate heat. Therefore, the more optical processing we incorporate into the system, the more heat needs to remove. And as a result, we can only employ optical infrastructure inside a data center.

The last one is the requirement for a direct band gap semiconductor in optical electronics to convert electricity into light. The goal is to use the electrons to promote the production of a photon. Because silicon lacks a straight band gap semiconductor, it cannot produce lasers. Therefore, none of this can occur in silicon; instead, you always need a device integrated back into silicon, amplifiers, or other devices. Thus, diverse integration and packaging make it difficult to advance this kind of technology.

What are some of the best optoelectronic technologies that have been used in operations?

Onboard optics is currently a category of modules. People are attempting to develop a module that can be simply plugged into a board, like other components added to boards in a standard SMT line, for communications and data centers applications. The materials you choose must be strong enough to withstand an SMT line. When you mount these onboard optical modules onto a PCB board, the optical portion of them must couple into a wave path that would be integrated into the board. As a result, the board would contain a transparent optical trace instead of a PCB copper trace, and there is where the light couples in. However, that technology can provide scale and compact size as well as ease of integration. Therefore, that has been under development at the industrial level for a while.

The second one we refer to as expanded beam technology between the lasers and the fiber. Linking a fiber that is emitting light or has a light but may only be 10 microns in size is one of the difficulties. Controlling the light's coupling into something else 10 microns in size is necessary.

It is now much simpler mechanically to join those two parts together thanks to connectors and lasers that can stretch that beam beyond 10 microns.

Could you shed some light on the effectiveness of embedding electronics and the development of sustainable energy technology?

As I mentioned before, the way heat is being produced by the increasing use of optics in data centers to address density. Data centers use a lot of electricity. Moving processes a little bit closer by employing optics makes them a little bit more efficient, and a lot of the world's grid power is actually used to power data centers. But it was with regard to solar cells. Solar energy capture with solar cells is a type of alternative energy. In other words, it is a detector that takes in sunlight and converts it into electricity. And the alloys used to make these solar cells are different from those used to make communications optics.

They keep improving in terms of efficiency. Therefore, they are still not entirely effective. However, modern solar cells and solar panels that you may install on your roof are significantly better compared to those that were available a generation ago. They are most likely orders of magnitude more efficient and less expensive. Therefore, the ability to increase optical efficiency—the ability to capture sunlight and convert it into electricity—is enabling that industry to do so. Also, it's allowing solar cells to proliferate on everyone's roofs or perhaps in new locations like car roofs or portable solar cells, etc.

How do you envision the future with all these potential disruptions and transformations that have happened over the years?

The march of speed and density has thus been the major development over that time. The first fiber optic underwater system launched at AT&T with a speed of 2.5 gigabits per second. That was a big phone that had a specific capacity for a number of phone calls, but not sure how many could be made at once. These systems currently use optics that is 100 gigabits in size, and they multiplex it even further so that you can transfer more than terabytes on one device.

"We created design guidelines for a certain fab, and now we have commercial fabs for the best electronics, where people adhere to their own sector's design guidelines"

So, efficiency is exploding. The ability of the internet to send all of this data globally is what actually makes it possible. The production processes have become considerably more under control when you ask me what's changed. In other words, when I started, it was more or less a cottage industry. There weren't really any design guidelines.

We created design guidelines for a certain fab, and now we have commercial fabs for the best electronics, where people adhere to their own sector's design guidelines. To encourage more suppliers to expand, you have standards-based products, Trans receivers, and other items. However, it still faces difficulties because it is a capital-intensive industry.

Therefore, if we could solve the coupling problem in a much more scalable and automated manner, optical devices would be adopted and used even more. However, I believe that's what's going to happen in the future. Because the dependability of the optics also needed to be established, there were these fairly standardized Trans receivers that allowed for plug-and-play in the center. Again, it took a lot of studies to demonstrate that lasers would last for 20 years before it was profitable to submerge them.

Next, they must either be able to be easily removed and replaced or survive long enough in a data center. Placing a device on a board requires considerably more care than placing a C'MO device when you're talking about having to couple tiny amounts of microns of light. It is, therefore, more difficult to get the entire system to accept optics as a method of data communication. Therefore, the robotics of light placement and coupling will enable future applications and the improvement of the devices themselves. They have thus become speedier.

Therefore, most of the silicon is needed for optoelectronics to drive signals at 100 Gbps and understand whether they are a zero or a one as they leave the detector. They must also advance in the world of C'MOs. Some of those are made of silicon germanium. Other technologies must therefore advance concurrently, but it has done so quite relentlessly. A factor of three orders of magnitude, a thousand or more, or both, is a lot in 30 years. When I consider that productivity, I believe it even surpasses what has occurred in the silicon and transition industry with Moore's law. However, there is still more to achieve in the optical domain.

What would be your advice to your fellow peers or aspiring professional?

Optics is quite a fascinating field where we met some fantastic people. We are capable of demonstrating amazing capabilities. The difficulty arises when attempting to scale it for a very high volume application because it is frequently capital demanding, and the methods required to produce a million of anything are frequently different from those required to produce just one. I believe optics presents a special challenge since sometimes the volume isn't at the front, and you need to invest money to establish the market. Therefore, my advice would be to truly consider the capital required if you're trying to develop a new application or scale up a new product.