BFR and the space industry

[john smith 19]

Which is the biggest flaw in current ideas of orbital or otherwise off-Earth manufacturing - no-one's found anything worth manufacturing! But then, you probably need to do a lot more experimentation in space before someone stumbles over such a product.

Not strictly true. There are a small number of products that are.

IIRC the top one is mfg a special kind of glass for fiber optic amplifiers, and it's pulling into those fibers.  The good news is such a process could be highly automated. The bad news is the very high power requirement it has for the furnace.

Of course if you had a colony - even one that just supports a tourist operation or science laboratories - then you'll likely develop local manufacture. A bit like the development of Las Vegas.

I'd wonder how much that had to do with it's proximity to various USAF bases and the Nevada Test Site. 
Otherwise Mars is a very long way to play Roulette or Blackjack 

[launchwatcher]

The ZBLAM fiber is worth from $100 to $10,000 per meter of length. A kg of source material will produce 6km of fiber or a produce value of $600,000/kg to $60,000,000/kg. Based on quality of fibre. It is expected that the space made fibers will be of exceptional quality therefore worth $10,000/m.

ZBLAM or ZBLAN?   

How much can you sell at $10/mm?  (The folks who build transoceanic cables will undoubtedly be able to negotiate a better price if it becomes available in multi-thousand-kilometer quantities..).

(There's also the risk that someone clever will come up with a manufacturing process that doesn't require microgravity).

[Robotbeat]

The ZBLAM fiber is worth from $100 to $10,000 per meter of length. A kg of source material will produce 6km of fiber or a produce value of $600,000/kg to $60,000,000/kg. Based on quality of fibre. It is expected that the space made fibers will be of exceptional quality therefore worth $10,000/m.

ZBLAM or ZBLAN?   

How much can you sell at $10/mm?  (The folks who build transoceanic cables will undoubtedly be able to negotiate a better price if it becomes available in multi-thousand-kilometer quantities..).

(There's also the risk that someone clever will come up with a manufacturing process that doesn't require microgravity).

Your last point is why it's critical for prices to orbit to come down. There will always be a lot of pressure to bring the process down to Earth when orbit costs $10,000/kg. But if it's just &10/kg, then why even bother?

[freddo411]

Orbital manufacturing isn't going to require factory workers. We are phasing out manuel labour on Earth. Sending hundreds of factory workers to space and keeping them alive is a huge expense that simply doesn't make any economical sense. If we ever see orbital factories, they will be as automated as possible.

You might not even need a factory. Just stuff the manufacturing equipment and raw materiel into a BFS, launch, produce a batch of your zero-g gizmos, and return the whole thing for maintenance.

All we need is to figure out what gizmos will economically benefit from being made in zero-g.

Anything that prefers a clean room or vacuum chamber on here on Earth.   Chip fab?   Carbon fiber autoclaves?

[JamesH65]

Orbital manufacturing isn't going to require factory workers. We are phasing out manuel labour on Earth. Sending hundreds of factory workers to space and keeping them alive is a huge expense that simply doesn't make any economical sense. If we ever see orbital factories, they will be as automated as possible.

You might not even need a factory. Just stuff the manufacturing equipment and raw materiel into a BFS, launch, produce a batch of your zero-g gizmos, and return the whole thing for maintenance.

All we need is to figure out what gizmos will economically benefit from being made in zero-g.

Anything that prefers a clean room or vacuum chamber on here on Earth.   Chip fab?   Carbon fiber autoclaves?

I think TSMC's latest fab is going to cost $40B to construct IIRC....anything that makes the chip biz cheaper will be of great benefit. Although I suspect you won't save a lot of that 40B by doing it in space.

[Semmel]

I think one of the limiting factors for chip production is the size of the waver. The waver is a cut from a mono-cristaline silicon structure, which is very hard to get at the sizes that are currently used. I am not a materials scientist but maybe someone can jump in and confirm or debunk the hypothesis that these crystals would be easier to create in microgravity?

[chipguy]

Orbital manufacturing isn't going to require factory workers. We are phasing out manuel labour on Earth. Sending hundreds of factory workers to space and keeping them alive is a huge expense that simply doesn't make any economical sense. If we ever see orbital factories, they will be as automated as possible.

You might not even need a factory. Just stuff the manufacturing equipment and raw materiel into a BFS, launch, produce a batch of your zero-g gizmos, and return the whole thing for maintenance.

All we need is to figure out what gizmos will economically benefit from being made in zero-g.

Anything that prefers a clean room or vacuum chamber on here on Earth.   Chip fab?   Carbon fiber autoclaves?

I think TSMC's latest fab is going to cost $40B to construct IIRC....anything that makes the chip biz cheaper will be of great benefit. Although I suspect you won't save a lot of that 40B by doing it in space.

The majority of the cost are the state of the art photolithographic tools going into that fab.

Absolutely nothing about space makes wafer fab easier or cost less. Quite the opposite actually.

In fact, IMO the very last thing an otherwise self-sufficient Mars colony will need to import from Earth are high end ICs.

[speedevil]

In fact, IMO the very last thing an otherwise self-sufficient Mars colony will need to import from Earth are high end ICs.

Perhaps not quite the very last - engineered viral, bacterial and fungal cultures can be some orders of magnitude more expensive than the highest end IC, and replicate so the value can be ridiculously more.

Seeds too perhaps.

[ZachF]

Orbital manufacturing isn't going to require factory workers. We are phasing out manuel labour on Earth. Sending hundreds of factory workers to space and keeping them alive is a huge expense that simply doesn't make any economical sense. If we ever see orbital factories, they will be as automated as possible.

You might not even need a factory. Just stuff the manufacturing equipment and raw materiel into a BFS, launch, produce a batch of your zero-g gizmos, and return the whole thing for maintenance.

All we need is to figure out what gizmos will economically benefit from being made in zero-g.

Anything that prefers a clean room or vacuum chamber on here on Earth.   Chip fab?   Carbon fiber autoclaves?

I think TSMC's latest fab is going to cost $40B to construct IIRC....anything that makes the chip biz cheaper will be of great benefit. Although I suspect you won't save a lot of that 40B by doing it in space.

Manufacturing in space could be a good way to protect IP though.

[IainMcClatchie]

I think one of the limiting factors for chip production is the size of the waver. The waver is a cut from a mono-cristaline silicon structure, which is very hard to get at the sizes that are currently used. I am not a materials scientist but maybe someone can jump in and confirm or debunk the hypothesis that these crystals would be easier to create in microgravity?

My dad used to have the end of a germanium ingot on his desk, from the 1960s.  2 inches in diameter.

Last year in the lobby at SunEdison I saw a monocrystalline silicon ingot.  12 inches in diameter.  20 feet long.

Photovoltaic manufacturers used to build their modules from scrap silicon wafers that had been rejected by the integrated circuit manufacturing industry.  Solyndra was started at a time when PV manufacturing had soaked up all the scrap, and was starting to buy their own wafers.  At the time, it seemed like the transition to paying full freight for the silicon was going to be a problem.  Solyndra was going to make a focussing system to get more out of each square cm of silicon.

So then a bunch of Chinese silicon foundries started cranking out monocrystalline silicon ingots specifically for PV manufacture.  They were below the cost of scrap IC wafers.  Nowadays the dominant use of silicon wafers is for PV and not for integrated circuits.  The tail and the dog have swapped roles.  Solyndra bet on the wrong technological fork and died.

[chipguy]

In fact, IMO the very last thing an otherwise self-sufficient Mars colony will need to import from Earth are high end ICs.

Perhaps not quite the very last - engineered viral, bacterial and fungal cultures can be some orders of magnitude more expensive than the highest end IC, and replicate so the value can be ridiculously more.

Seeds too perhaps.

I am not saying other items can't be more valuable either absolutely or in value per unit mass.

The key point is the manufacturing infrastructure for state of the art semiconductors is the most difficult to duplicate and bring up on another planet. Gene sequencing and splicing equipment is making huge strides right now in miniaturization and cost reduction. R&D and production scale facilities for engineered biologicals are already much smaller and far less expensive than a wafer fab for leading edge process technology and this gap will likely grow rather than shrink. Then there is the question of secondary infrastructure for production of ultra-pure consumables for a wafer fab. Biological processes are far more forgiving of input quality.

[envy887]

Convection (which micro-g solves) doesn't seem to be an issue for growing defect-free silicon crystals. But does seem to be an issue for growing some other potentially useful crystals like silicon carbide.

[oldAtlas_Eguy]

The ZBLAM fiber is worth from $100 to $10,000 per meter of length. A kg of source material will produce 6km of fiber or a produce value of $600,000/kg to $60,000,000/kg. Based on quality of fibre. It is expected that the space made fibers will be of exceptional quality therefore worth $10,000/m.

ZBLAM or ZBLAN?   

How much can you sell at $10/mm?  (The folks who build transoceanic cables will undoubtedly be able to negotiate a better price if it becomes available in multi-thousand-kilometer quantities..).

(There's also the risk that someone clever will come up with a manufacturing process that doesn't require microgravity).

Your last point is why it's critical for prices to orbit to come down. There will always be a lot of pressure to bring the process down to Earth when orbit costs $10,000/kg. But if it's just &10/kg, then why even bother?

Sorry your are correct it is ZBLAN.

But the sense I get from some of the comments by MIS and their co-sponser for this project who is the current primary manufacturer of ZBLAN is that to be able to manufacture exceptional quality ZBLAN at the same price as the current low quality ZBLAN. What this would do is cause an explosion in the usage of ZBLAN and creating much more revenue and profits than what they are experiencing currently. As long as the quality and difficulty remains as it does now the market for ZBLAN will struggle. But lower the price and it would boom.

But the sad part here is that this is but a single very special case. Other cases of similar nature will likely be rare. But if the cost of travel to and from space becomes little more that the cost of travel to another continent then very marginal cases and complete impossible cases become a competitive business solution. That is what BFR very low prices will do for Space Industry.

[launchwatcher]

I think one of the limiting factors for chip production is the size of the waver. The waver is a cut from a mono-cristaline silicon structure, which is very hard to get at the sizes that are currently used. I am not a materials scientist but maybe someone can jump in and confirm or debunk the hypothesis that these crystals would be easier to create in microgravity?

My dad used to have the end of a germanium ingot on his desk, from the 1960s.  2 inches in diameter.

Last year in the lobby at SunEdison I saw a monocrystalline silicon ingot.  12 inches in diameter.  20 feet long.

My understanding is that most of the big fabs use 300mm (~12 inch) wafers these days but 200mm is still used.   An industry push to move to 450mm a few years back appears to have stalled; the increased weight of larger-diameter ingots and wafers was cited as one of the factors in the news reports I read.

Round wafers are carved into rectangular chips.   The main motivation for larger-diameter wafers is less wastage around the edge when you slice a circular wafer into rectangular chips.    The growth in chip area seems to have slowed down which reduces the urgency of moving to bigger wafers.

[Semmel]

I think one of the limiting factors for chip production is the size of the waver. The waver is a cut from a mono-cristaline silicon structure, which is very hard to get at the sizes that are currently used. I am not a materials scientist but maybe someone can jump in and confirm or debunk the hypothesis that these crystals would be easier to create in microgravity?

My dad used to have the end of a germanium ingot on his desk, from the 1960s.  2 inches in diameter.

Last year in the lobby at SunEdison I saw a monocrystalline silicon ingot.  12 inches in diameter.  20 feet long.

My understanding is that most of the big fabs use 300mm (~12 inch) wafers these days but 200mm is still used.   An industry push to move to 450mm a few years back appears to have stalled; the increased weight of larger-diameter ingots and wafers was cited as one of the factors in the news reports I read.

Round wafers are carved into rectangular chips.   The main motivation for larger-diameter wafers is less wastage around the edge when you slice a circular wafer into rectangular chips.    The growth in chip area seems to have slowed down which reduces the urgency of moving to bigger wafers.

I am not sure that the waste at the edge is the driver. The machine that imprints (to keep it simple, ok?) the chip layout onto the wafer is enormous and using it expensive. Exchanging wafers is a large waste of time for the machine so you want as large wafers as possible. It boils down to: If you had larger wafers, chip production would be cheaper.

So if you could make wafer ingots in space that have double the diameter than on earth, you could practically cut the price of one chip by a factor of 4. And if the price for the crystal ingot was double or triple the original price, it wouldn't even matter. Not sure if that is a good motivation to go to space though. After all, a wafer factory in space and servicing it would be mighty expensive as well.

@edit: see italic.

[IainMcClatchie]

I hadn't heard that the 450mm transition had stalled.  I was surprised that the solar industry was still at 300mm diameter -- they'd be the obvious folks to lead the way to larger diameters, as the rectangles they cut out of those things are far bigger than the IC folks, and they just produce much more area now.

BTW, Semmel, the critical machine that prints the various layers is called a stepper.  It doesn't image the entire wafer in one go, but rather has a reticle size, usually around 20 mm diameter.  It prints that, moves over, and repeats.  Larger wafers take more time.  There is a little savings with big wafers because you spend less time moving it in and out of the stepper.  I think the big savings are all the whole-wafer operations, like spinning on photomask and etching.  PV manufacturing doesn't need a stepper.

I think the idea of manufacturing in space is just bonkers.  The future seems pretty obvious to me.  It's going to be LEO comsat constellations.  There will be three, SpaceX, OneWeb, and a Chinese one.  Once the first generation shows profitability, the second generation will grab most of the internet and cellphone market.  Launch vehicles will get optimized for putting this second generation in orbit, and everything else will make do with those vehicles.  Things will be pretty good -- reusable two stage launchers with big fairings for fluffy payloads with lots of antennae and solar arrays and thermal radiators.  The BFR is a fine idea, the BFS is daft, SpaceX will either go with a simple reusable second stage by themselves, or Blue Origin will show them how to do it.

Bigelow will be happy, NASA will be happy, Elon will probably be frustrated, and folks with ground-based telescopes will be annoyed at the ubiquitous antennae flares.

[192]

  The BFR is a fine idea, the BFS is daft, SpaceX will either go with a simple reusable second stage by themselves, or Blue Origin will show them how to do it.

I'm curious as to what you think is daft about the BFS, or at least the satellite launch variant, and what you think a "simple reusable second stage" looks like in comparison.

You say the BFR is fine, by which I presume you mean the booster, so I take it you don't object to the scale/ carbon fibre tanks/ raptor.

Other than those the main differences between BFS and a conventional second stage: are an integrated fairing, a heat-shield, winglets, landing engines and legs. Which of these do you object to and why? and how would you do it differently?

Most of these are a necessary part of their reentry and landing strategy. The only one that isn't is the integrated fairing, but if you're going fully reusable you want it back, and having one reentry solution in place of two seems like the simpler solution to me. Yes it has a mass penalty by taking the fairing all the way to orbit, but it provides extra surface area and an aerodynamic nose for reentry, it simplifies recovery and reuse operations, and the mass penalty is less important on a fully reusable vehicle. 

[IainMcClatchie]

Well said.

I'm especially fine with the scale.  They are going to need gobs of lift for the comsat constellation.  Elon is right on point when he says that chartering a 747 from North America to Australia, and back, costs about what a new Cessna 206 goes for.  A 747 is 220 tonnes.  Last year's iteration of BFR was penciled at 275 tonnes, this year it's smaller.  It'll be an evolutionary step for building large composite airframes.

I'm a little less impressed with the switch from 12m to 9m.  Rockets don't scale well with height.  Essentially you are lifting a tank of LOX on a base of compressed gas.  If that column gets taller the average pressure at the base must get larger.  Booster engine exit pressure wants to be around 40-50 kPa, and average base pressure can increase with increasing chamber pressure... but it's not 1:1.  So once the rocket is as tall as the Saturn V, stop scaling height and scale diameter instead.  Also, SpaceX should budget for lower density payloads, which implies larger diameter fairings.  The 3.6m road transport standard was a smart move, but once they commit to barging it they should go big and squat.

Unnecessary problems with BFS:

Integrated Fairing  The payload is bigger in volume than the upper stage.  That fairing is huge and heavy (the F9 fairing is about the same mass as the second stage!) and as you say is expensive to take to orbital velocity and back.  I don't think it should come down on its own, because then you need ships out there to get it regardless of where the booster is landing.  It should be a part of the BFR and land with that.  This means it encapsulates the second stage as well as the payload, but that's got benefits as well as costs.

Heat Shield  PICA-X is great stuff, but like the Shuttle tiles it's porous.  The BFS design puts it on the outside of the spacecraft to get rained on while on the pad.  The absorbed water will freeze while in orbit and can break the tiles, and can also break tiles while vaporizing during reentry.  The Shuttle had lots of delays and work associated with its exposed heat shield.

The obvious place for the heat shield is inside the fairing.  It'll still need refurb because it gets exposed to moisture after landing, but for high-tempo operations they might conceivably get the PICA inside a controlled low-humidity atmosphere before it gets cold.

Legs  They will have the technology to land on a cradle.  I'm not as exercised about this as the fairing, as the legs are 1/6th as much mass.

People  Since the vast majority of upmass in ten years will be comsats, people and their related gear should go in something that doesn't affect the comsat launching missions.  If I was going to Mars, I'd want to see something that had an inflatable part that aerobrakes but doesn't reenter, and maybe large propellant tanks that get filled while in orbit around Mars.

[Asteroza]

I think one of the limiting factors for chip production is the size of the waver. The waver is a cut from a mono-cristaline silicon structure, which is very hard to get at the sizes that are currently used. I am not a materials scientist but maybe someone can jump in and confirm or debunk the hypothesis that these crystals would be easier to create in microgravity?

My dad used to have the end of a germanium ingot on his desk, from the 1960s.  2 inches in diameter.

Last year in the lobby at SunEdison I saw a monocrystalline silicon ingot.  12 inches in diameter.  20 feet long.

My understanding is that most of the big fabs use 300mm (~12 inch) wafers these days but 200mm is still used.   An industry push to move to 450mm a few years back appears to have stalled; the increased weight of larger-diameter ingots and wafers was cited as one of the factors in the news reports I read.

Round wafers are carved into rectangular chips.   The main motivation for larger-diameter wafers is less wastage around the edge when you slice a circular wafer into rectangular chips.    The growth in chip area seems to have slowed down which reduces the urgency of moving to bigger wafers.

I am not sure that the waste at the edge is the driver. The machine that imprints (to keep it simple, ok?) the chip layout onto the wafer is enormous and using it expensive. Exchanging wafers is a large waste of time for the machine so you want as large wafers as possible. It boils down to: If you had larger wafers, chip production would be cheaper.

So if you could make wafers in space that have double the diameter than on earth, you could practically cut the price of one chip by a factor of 4. And if the price for the crystal ingot was double or triple the original price, it wouldn't even matter. Not sure if that is a good motivation to go to space though. After all, a wafer factory in space and servicing it would be mighty expensive as well.

Making wafer ingots in space strikes me as more viable due to return packaging issues (easier to carry a single ingot than lots of wafers back to earth).

I believe it was mentioned somewhere that the new Nvidia GPU's are maxing out the reticle size of the steppers at TSMC, which are around 80mm. Size of reticle up to a point improves throughput at higher wafer diameter.

[IainMcClatchie]

I believe it was mentioned somewhere that the new Nvidia GPU's are maxing out the reticle size of the steppers at TSMC, which are around 80mm. Size of reticle up to a point improves throughput at higher wafer diameter.

No.

NVidia's largest GPU is the Tesla V100 and is 815 mm^2.   It sits on top of a 1200 mm^2 silicon interposer.  TSMC has said it uses reticle stitching for the interposer, and it's possible they are stitching the V100 as well.  If not, it would require a 40mm reticle diameter, which is astonishing.  Among other things, such an enormous reticle would require a new, larger and higher performance lens at this process node.  These lenses take years to build and are a significant portion of the R&D effort for a new process node.  That is, can it be done at all.  There is no way they'd build a custom lens for even 10% of TSMC's fab output.  The V100 is maybe 0.0001% at best.

Below is an image of a stepper.  I swear, Lars, that I will not respond to another post not having to do with rockets.