This page contains a Flash digital edition of a book.
epitaxial layer structure upside-down onto the processed silicon wafer. Then the InP substrate is removed. What remains is thin, III-V epitaxial layers bonded onto the silicon circuitry, which, since it is a planar wafer at that point, can be processed in a full wafer fab. The processing of the laser is carried out at the wafer scale, but the actual bonding step itself is a multiple die-to-wafer attachment process. ‘Several groups in the world have

mastered this technology quite well and have integrated lasers on silicon with respectable performance,’ commented Baets. ‘However, the technology so far hasn’t been brought to industrial maturity. Intel has gone a long way in doing that. But still there are hardly any CMOS fabs in the world where you could have these integrated lasers being processed on the silicon at the moment.’ There is also a third option, which is still much more at a research stage and far from maturity, but if it materialises then it might prove to be the winner in the long run, according to Baets. This is to grow III-V semiconductor material directly onto silicon – not bonded, but epitaxially grown. This has been notably difficult, because there’s a large crystal lattice size mismatch between silicon and, for example, InP, along with various other issues. Therefore, in the past, growing InP directly on silicon resulted in very poor-quality material that was full of defects. But in recent years

there have been several groups reporting successes in reaching good quality crystalline material, according to Baets. ‘There is an increasing interest in direct growth of III-V on silicon – and, if that is possible, then you have a potential route towards full wafer technology in a CMOS fab including the laser. It’s something not yet mature, but there are optimistic signs in recent years that it might be realistic.’ IBM has an interest in silicon photonics for data centre applications and high-performance computing. ‘For IBM, big data is a very important aspect. Here, high-bandwidth, efficient communication within and between data centres is important,’ commented Offrein. IBM’s silicon photonics platform combines both electrical and optical functionality, which, according to Offrein, is different to what many others are doing in this area. ‘We want to bring the critical electrical circuitry very close to the optical devices they are driving,’ said | @electrooptics

Offrein – so the driver circuitry is physically close to the electro-optical modulators and the amplifiers are close to the germanium detectors. ‘Such a short distance means you have optimum performance, and we can design the components in such a way to have an ideal match. ‘What we’re [IBM] exploring in research is: what is a reasonable path in order to move towards tighter integration of the laser on silicon? Today, it’s a separate laser, but we see large potential in bringing the laser onto the chip,’ Offrein commented.

It used to be that CMOS fabs would regard any form of III-V material as an impurity, but as they started moving towards ever smaller nodes, like 14nm and 8nm, III-V began to be needed to make higher-performance transistors. Dr Jens Hofrichter, a postdoctoral researcher at IBM Research, Zurich, noted that, by building on this, the CMOS community can start thinking about integrating lasers onto silicon photonics using the same technologies and methods.

IBM has an

interest in silicon photonics for data centre applications and HPC

Letting the heat out The light source challenge is complicated further by the temperatures the lasers would have to operate at as part of a data centre application, for instance. Dr Dennis Deppe, a professor at CREOL, the College of Optics and Photonics at the University of Central Florida gave a presentation at the beginning of May at IEEE’s Optical Interconnects Conference in California, USA, titled: ‘Letting the heat out: silicon photonics

integration with VCSELs’.

Dr Deppe is working on integrating vertical- cavity surface-emitting lasers (VCSELs) on silicon chips. ‘In some ways VCSELs are viewed as competition for silicon photonics,’ Deppe told Electro Optics. ‘I view them as a laser device for silicon photonics, because I see advantages in integrating those lasers with silicon electronic circuitry.’ According to Deppe, most of the big applications for silicon photonics, such as data centres and high-performance computing require high- temperature operation. ‘The laser has been a big barrier, because a lot of the lasers that are being implemented and the ways that they’re being implemented don’t allow the lasers to work at high temperatures,’ he said. There has been previous work integrating


Picosecond Timing ∆ t1

∆ t2


stop t [ps]


Now with USB 3.0

HydraHarp 400 •

• •

• • •

• •

Up to 8 synchronized but independent channels and common sync Versatile measurement modes High data throughput


Coincidence correlation Event timing in LIDAR Time-resolved fl uorescence spectroscopy Quantum optics Optical tomography

PicoQuant GmbH Berlin, Germany

Meet us at booth #1727

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45