Skip to main content

MASS manufacturing of TrAnsceiveRs for Terabit/s era

Periodic Reporting for period 1 - MASSTART (MASS manufacturing of TrAnsceiveRs for Terabit/s era)

Reporting period: 2019-01-01 to 2020-06-30

To address the next generation of transceivers in Data Centre (DC) infrastructures, targeting 800G and >1Tb/s aggregate data rates, expected to massively use Silicon Photonics based PICs, not only fabrication processes needs to be enhanced, but also the entire module manufacturing process, including packaging steps, and test and measurement operations. Indeed, with increasing manufacturing volumes, and increasing complexity level of the Silicon Photonics PICs, existing technologies are too limited in term of scalability, manufacturing throughput and test duration. As an example, today, assembling a ribbon fiber, held in a glass v-groove array, to a PIC through active alignment is typically limited by:
* The minimum pitch of fiber I/Os is currently at 127µm, an obstacle that severely limits the optical I/O density of the PICs. Additionally, this value is higher than the typical pitch value of pins in electrical circuits that questions one of the main advantages of photonics; small pitch value due to the absence of electromagnetic crosstalk.
* The assembly process time, typically 10 - 20 minutes including changing the parts in a state of the art 6-axis alignment robot, performing a passive pre-alignment, an active final alignment, application and curing of epoxy and placing back the assembled transceiver to the shipping box.

Moving to the next generation of devices requires the development of new approaches, while the expected time to market limits the introduction of breakthrough solutions e.g. new materials like polymers or full passive alignment strategies that are not compatible with the required reliability level, nor compatible with standard semiconductor assembly processes. However, a number of established technologies are mature enough and exist in Europe that can be wisely combined in order to address the Terabit/s integration challenge, so as to drastically reduce the €/Gb/s figure of merit and break the throughput barrier due to legacy technologies limitations.

MASSTART targets the exploitation of the following technologies to achieve the targets of enhanced performance at the lowest possible cost for the next generation 800Gb/s and higher Datacom transceivers :
* Glass waveguide technologies, giving access to spot size conversion and pitch conversion to address the need of fiber coupling with higher densities (down to 15µm pitch).
* Laser integration, using a priori good bare dies combined with state of the art MEMS type submounts or glass interposers. This topic still remains a concern for high yield manufacturing of photonic modules, as most of the wafer level assembly techniques remains at TRL less than 6 for now.
* Assembly techniques, relying on a mix of active alignment and pattern recognition, that will reduce the assembly time by a factor of x6 and improve the throughput to 30 pieces/hour from the current 6 pieces/hour with standard technology
* Characterization of the high speed modules at wafer level with a with new set of tools that can reduce the overall time by a factor of x10, down to 1 minute per wafer.
* Photonic/Electronic Integration, leveraging last decade development in the field of 3D packaging, especially flip-chip bonding at low pitch (40µm and less) and Through Silicon Vias (TSV), enabling dense interconnect of high IO count PICs to BGA substrates.

In addition to that, renowned end-users (module and network building blocks providers) are willing to lead developments, following a roadmap to On-Board module integration . Coordinating all these skills and know how, especially those related to assembly and test automation, at the European level within the MASSTART project, will allow Europe to unleash a breakthrough in Photonic transceivers assembly targeting data rates of at least to 400Gb/s aggregate bandwidth, in a relatively short term of 3-5 years. The mass automation in the assembly and the characterization is expected to bring down the cost for the rack-to-rack communication links to less than €1/Gb/s with minimum intervention by human hands. By addressing this major cost problem, MASSTART will reverse the current trend of job migration to Asia for exploitation of cheap labour and will render feasible the repatriation of many photonic jobs back to Europe.

The development of the MASSTART industrial and self-consistent process flow will be assessed by fabricating and characterizing demonstrators, addressing the mid-term requirements of next generation transceivers required by DC operators and covering both inter- and intra- DC applications. It will be devoted to help a standardization for future modules and their related packaging.
The MASSTART project intends to establish a new assembly and test paradigm for next generation 800G and 1.6 T transceivers, including On-Board configuration, using robust, low cost, and high throughput packaging and test techniques, thus increasing Europe mass manufacturing capability for Datacom modules.
The first objective of the MASSTART project is to standardize a set of glass waveguide based interfaces allowing high density PIC/fiber interconnects and laser/PIC coupling (loss<0.5dB pitch 15µm).

The second objective of the MASSTART project is to develop a micro submount, with the related assembly process, to allow laser hybrid integration onto the PIC with alignment tolerances <1µm and angular error<0.5°.

The third objective of the MASSTART project is to develop an optimized equipment and process for high throughput (improvement factor of 6) assembly of the aforementioned part together with the PIC.

The fourth objective of the MASSTART project is to develop a full PIC fabrication process flow embedding TSVs (transmission loss lower than 2dB at 60GHz, pitch 40µm).

The fifth objective of the MASSTART project is to setup test methodologies at wafer level for reduced time during characterization (improvement factor of 10 in characterisation in time) and yield analysis (targeted yield 90%).

The sixth objective of the MASSTART project is to enhance the capabilities of flip chip bonders for enhanced performance and reduced assembly time by factor of 6.

The seventh objective of the MASSTART project is to establish a standard process flow for Terabit/s class transceiver modules assembly and test.

The eight objective of the MASSTART project is to assess the full assembly & test flow by demonstrating two type of terabit class transceivers (C-Band 600Gb/s coherent for inter data center, and O-Band 1.6 Tb/s on-board for intra data center).

The ninth objective of the MASSTART project is to manage a standardization activity for next generation Terabit/s transceivers, and the related packaging.
09_Credits Acknowledgement
01_rise up to the currents waves
05_novel applications create more traffic
08_MASSTART transceiver specs
07_MASSTART partners
06_MASSTART goal
Clean room
03_data is collected in immense data centers
02_data is everywhere
04_transpoted processed and analysed