Finishing up our look at the Suss Technology workshop at SEMICON West.
Nanium
Ricardo Gai, head of eWLB engineering at Nanium gave a presentation on wafer level fan out packaging, FO WLP. Nanium, headquartered in Portugal, is derived from the following bloodlines:
Nanium has been in 300mm HVM of FO WLP since Q3 2010 and has shipped more than 350MM packages. For a complete discussion of eWLB see IFTLE 124, “Status and the Future of eWLB.”
Process Challenges with eWLB:
- Warpage
– Warpage and shape changes during processing; stiffness is a key parameter to control;
– Equipment modifications are required to handle FO-WLP wafers
– hardware – new end effectors and new chuck sealing
– software – upgrades for warpage handling
- Die-Shift – In case of complex devices like (Multichip SiP), mask aligners are not capable of overlay requirements and a stepper is required.
Multi-Layer RDL for multilayer RDL the overlay must be smaller than the die shift. This also requires stepper technology.
Fine RDL Line Width / Space and Fine Ball Pitch – – better resist technology is necessary to achieve L/S = 10/10
- Cost Pressures
– more die per carrier by using smaller edge exclusion zone
– larger substrate sizes
– 300mm and large panel processing to reduce costs.
- Future Challenges:
– thinner substrates will require temp bonding and front side protection
GlobalFoundries
Jon Greenwood of GlobalFoundries gave a presentation on 2.5/3D readiness for HVM. Global Foundries is dedicated to support the Collaborative Partner Model:
They are creating a supply chain where GF is responsible for the wafer processing and backside integration (BSI) and assembly is owned by the OSAT partners.
As we have discussed before [see IFTLE 142, “GlobalFoundries2.5 / 3D at 20nm…” ] 2.5D interposer work is ongoing in Singapore and 3D activity is in NY. In June 2013, they announced a certified set of design flows for 2.5D interposers. As the next slide shows, by 2015 they hope to offer TSV “in all nodes.”
Greenwood announced that GF thinks their 2.5D and 3D packaging capabilities are proven and their primary focus moving forward is yield and COO reduction.
Amkor Ron Huemoeller Sr VP of Advanced Product Development at Amkor presented their position on 3D/2.5D Market readiness.
Design capability status – established
– Mass production of optical sensors since 2008
– Worlds first fully integrated MCM TSV product started production in 2011
– System integrated architecture proven on many platforms such as:
– multiple logic on Si interposers
– logic + memory on Si interposers
– memory / memory stack
– memory / logic combination
Manfacturing Capability
– fine pitch Cu pillar in HVM
– wafer thinning equipment and infrastructure in place
– TSV etch equipment and infrastructure well-established
– Backside passivation – equipment and infrastructure well-established
– Backside bumping – equipment and infrastructure well-established
– wafer support – equipment and infrastructure in place but still needs improvement
– Assembly can be done chip-on-substrate, chip-on-wafer or chip-on-chip
– more work is needed on testing memory prior to committing to package stack.
Memory
– End customer chooses memory supplier (only primary memory sources today)
– Receive as KGM on tape and reel
– receiving 2, 4 die stacks in wide IO format
Amkor now sees interposer use reaching “value markets” post 2015.
For all the latest in 3DIC and advanced packaging, stay linked to IFTLE.