Apple’s 2020 iPhone Will Reportedly Feature 5G, 5nm Lithography

Apple’s 2020 iPhone Will Reportedly Feature 5G, 5nm Lithography

Apple wants the 2020 iPhone to make a splash, and the company has reportedly pulled out all the stops to make it happen. We’ve heard reports to this effect for months now — even before the iPhone 11 launched, there were indications that the Cupertino company wanted a big follow-up for this year’s device. Next year we’ll supposedly see a full range of 5G-capable iPhones with SoC’s built on TSMC’s 5nm lithography node.

Apple is already a bit behind the Android world as far as 5G introductions are concerned, but that hasn’t been much of a factor in 2019. While you can buy a 5G device today, network support for them is so scattered, it makes little sense to do so. Outside the US, the situation is a little different — there’s more 5G service available, and the spectrum being used is different, providing better range and support. 5G will be ramping up in the US throughout 2020, but availability will often be limited to downtown in major cities.

If Nikkei is right, Apple will put a major push on 5G next year, with three new devices all equipped with the latest standard. “It will be the first time Apple introduces 5G iPhones … There will be three of them and the company has set an aggressive sales target,” one of the people familiar with the company’s thinking said. The company wants to ship 80 million 5G devices next year. There are estimates that manufacturers will sell ~206M 5G handsets next year, about 18 percent of the expected yearly total.

How many of those users will actually have access to a 5G network? Considerably fewer — but provided the device provides significant benefits on its own, that may not matter much. 5G’s impact on battery life is uncertain. Current reports suggest that the density of base station requirements will drive higher power requirements partly offset by lower power consumption per base station. In smartphones, 5G uses less power when delivering high data bandwidth but more power than LTE when serving smaller amounts of data. Firing up the entire 5G modem for a relatively low data stream is not power-efficient.

Supposedly the new iPhone will use Qualcomm’s new 5G modem, the X55. While Apple bought Intel’s 5G business earlier this year, there’s no chance of the company having new silicon ready to go this quickly. Intel’s inability to ship that silicon to meet Apple deadlines are the reason why Apple bought the business unit and paid Qualcomm billions of dollars for a modem license. The deal between the two companies put an end to Apple’s antitrust lawsuit against Qualcomm — a lawsuit Cupertino had precious little reason to settle under other circumstances.

Besides this, the new, expected A14 will be made on TSMC’s 5nm process node. If you’ve paid attention to EUV (Extreme Ultraviolet Lithography) development, you’re aware that 5nm is a major node transition for TSMC. While EUV is being deployed on 7nm, it’ll only be used for limited steps at that node — contacts and vias only. At 5nm, TSMC will introduce EUV in more steps. This has been believed to require the development of a pellicle — a protective layer that sits over the wafer and prevents debris from landing on it, damaging the manufacturing process.

Prototype EUV pellicle. Image by ASML.
Prototype EUV pellicle. Image by ASML.

There have been some positive developments on the pellicle front. According to lithography expert Dr. Christopher Mack, Intel has presented data at the industry SPIE conference showing that if your die size is small enough, manufacturing without a pellicle can still make sense. It depends on how clean you can keep the scanner and what your losses are from particle adders relative to the loss of scanner throughput caused by introducing a pellicle. He writes:

When there are many die on one reticle, living without a pellicle may make sense. In this scenario, wafer (print die) inspections are used to find reticle defects (“repeaters” on the wafer) after they occur. Despite efforts to keep the inside of the EUV scanner clean, 20 percent of the time reticles at Intel developed particle adders between inspections, killing yield for those die. On the other hand, the use of a pellicle, which today has a one-pass transmission of 83 percent, results in a 30 percent loss of scanner throughput. (Note that using a pellicle also requires the use of a membrane just above the wafer to block out-of-band radiation, and this membrane has about 90 percent transmission. Overall light intensity is reduced by 0.9*0.83*0.83.) Which is more expensive, the lost scanner throughput due to low pellicle transmittance, or the lost yield due to printing reticle defects? That will depend on the die size.

We don’t know yet if TSMC will have a pellicle solution ready for Apple’s 5nm process specifically, or if they’ll try for non-pellicle manufacturing. The point is that there are potential solutions in play whether a pellicle is ready or not.

According to TSMC, the 5nm node offers relatively small improvements on power efficiency and performance. Gains on 5nm are expected to be up to 20 percent reduced power, up to 15 percent higher performance, and up to 45 percent area reduction. That’s a side effect of how Moore’s Law scaling has broken down — we can still pack transistors in more tightly, but doing more with them is going to be difficult. On the other hand, it’ll be interesting to see if Apple’s SoCs emphasize shrink over performance or power.

Other features supposedly include a new 3D depth-sensing rear camera with the ability to detect objects in the environment for use in AR apps and games. It isn’t clear what sort of impact 5G will have on battery life, but watch for how it shapes up before new iPhones go on sale. If multiple devices are all using the X55 modem, we should be able to judge performance fairly well before it ships.

Continue reading

Chip-Making Companies Can’t Buy Enough Lithography Machines
Chip-Making Companies Can’t Buy Enough Lithography Machines

It won't do much good to build a new fab if you don't have any machines to put inside it.

Intel Reportedly Won’t Deploy EUV Lithography Until 2021
Intel Reportedly Won’t Deploy EUV Lithography Until 2021

Intel will reportedly delay its introduction of EUV until 2021, while TSMC and Samsung hope to have the tech online by 2019.