Comparison between InP and Silicon Photonics

The most important qualities when comparing InP and Silicon Photonics PICs are functionality, performance and cost. Functionality and performance: The fundamental and key advantage brought by InP is the ability to monolithically integrate compact and efficient optical amplifiers.

Optical amplifiers play a comparable role in photonics to that of transistors in electronics. They are at the heart of any laser or laser system, but they are also used in many other applications, like switching, broadcast, wavelength conversion, power level control etc.

Without efficient integration of optical amplifiers the road to large scale integration is blocked: it is impossible to provide large scale photonic ICs with sufficient amplifiers in a hybrid way. That’s why InP-based photonic ICs are clearly leading in circuit complexity.

For less complex chips the advantages of InPbased monolithic integration are also substantial. For active devices, like lasers and modulators, InP performs far better than silicon-based technologies despite recent and hard won advances in the field.

There are no practical solutions on the horizon for efficient monolithic integration of lasers and optical amplifiers. Instead much research effort has focused on hybrid integration of InP components onto silicon, which comes with significant coupling losses. This has important consequences for optical power budgets and energy efficiency.

For the cost of integrated photonics solutions it is important to distinguish between chip cost and module cost. In many applications, particularly those requiring multiple optical and rf connections, the cost of the chip is only a small part of the total cost of a module, because the assembly and packaging costs are dominant. The technology that can integrate most functionality into a single chip will lead to the largest reduction of the costs of assembly and packaging and the associated loss in performance. So even if the InP chip is more expensive, the cost of the InP-based module will be lower, and its performance higher.

It is questionable, however, whether InP chips are really more expensive than silicon photonics chips. Here we have to distinguish between small to medium volume markets and high volume markets. From the analysis in the previous paragraphs we seefigure 5 that due to the large scale of its fabs and the associated high investments, silicon photonics is not well positioned to serve small and medium sized markets and InP-fabs have a better offer, both in terms of functionality and price.
The dashed line (200 mm fab) in the Figure (chip cost) can be considered representative for a small silicon photonics fab.

We see that volumes well over one million chips per year are required to make such a fab competitive with respect to chip cost, and for larger fabs even larger volumes are required.
For very large markets silicon photonics is better positioned than for smaller markets, but as can be seen from the Figure the same holds for InP: if the wafer size and the fab size is scaled up, the chip cost will go down in a similar way as for silicon photonics, whereas at the same time, the performance remains better. The contribution of the material costs (InP or Si) to the total chip cost is negligible. The chip cost is determined by the cost of the processing, and this is dependent on the size of the wafers and the size of the fab. A point of concern may be the wafer size, which is presently limited to 6” for InP. However, research is also going on for heterogeneous integration of a fully InP-based photonic layer on silicon (IMOS). We may speculate that such a technology could bridge the present gap between InP and silicon-based photonics, and may prove a faster track to wafer scale integration of photonics and electronics than via hybrid integration of InP and silicon photonics.

In summary: The wide-spread believe that InP photonics is much more expensive than silicon  photonics is highly erroneous. Not only does InP offer significant performance advantages, but, for the short and medium term, InP-based photonic integration offers clear cost advantages over silicon photonics as well.

More about the JePPIX roadmap.

2 Replies to “Comparison between InP and Silicon Photonics”

  1. Can I use this graph in a public presentation (OFC 2021 market Watch)
    I will give full reference and advertise the conference

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.