Molecular assemblers?

hammertime

What technological challenges stand in the way of making molecular assemblers? I mean, what exactly would it entail to generate some sort of nanotechnology that could assemble, from scratch, a car or something? I've heard something about the "fat fingers" and "sticky fingers" problem and something about a debate between Eric Drexler and a professor name Smalley.

I'm just wondering if it's possible that we'll, one day, have molecular assemblers. Any thoughts?

Ryan_m_b

This is a big question. Drexler and Smalley argued over the practicalities of building a molecular assembler based on the principles of mechanosynthesis. The idea behind this is that if you can build molecular sized grips to hold individual atoms you could possibly manipulate the molecules to break or make chemical bonds. With a well (and I mean bloody well) designed system it may be able to co-ordinate huge (so huge you need standard form to measure it) numbers of these actions at once in such a way as to break things down at an atomic level and rebuild them.

Now it could be argued that molecular assemblers do already exist. Life is a good example of something that breaks down macroscale objects into their microscopic components and rebuild those microscopic components back into desirable macroscale objects.

The issues with molecular assemblers that appear as desktop atomic factories are as follows;

1) Designing the thing -
Such a system would be hugely complex, it's analogous to trying to design a human being from scratch

2) Limitations of certain structures -
Complex chemicals are only stable depending on very specific environmental conditions (temperature, pressure, pH etc). Trying to make these structures could be like trying to build a car with play doh that keeps changing it's composition, material characteristics etc.

3) Dealing with wear and tear -
If you have a small error it's possible that the system will break, if your atomic tool-tip moves wrongly (perhaps because of temperature fluctuations) and binds a piece of machinery to a chemical it's dealing with then how do you fix it? You have to design the system to reabsorb and rebuild itself and design it so that this repair system is more efficient than the rate of damage. Considering the complexity of molecular repair faculties in biological systems we are back to square one.

To sum up I don't think there is a legitimate argument that such a thing could not exist however the complexities involved are fantastic. You have to design a system far more complex than all the life that has evolved on Earth. Now very limited "molecular assemblers" do already exist in industry and more may come (e.g. redesigning ribosomes) but a universal constructor as envisioned by some science fiction writers (and some popular scientists) are going to remain fantasy for a long time. The science and technology needed to build them is astounding.

hammertime

Well, what about at really cold temperatures? Wouldn't it be much easier to put all macroscale objects together at really (and I mean REALLY) cold temperatures?

Ryan_m_b

I going to take it that by this you are proposing the concept of a machine that's a cross between a 3d printer and an atomic force microscope. This solves some problems, doesn't change others and creates a whole load more.

At very cold temperatures materials are going to have different properties to what they would have at room temperature (or whatever temperature you intend for them to work at). This change in properties would be extremely detrimental when trying to thaw the devise out of the machine, a good analogy to this would be what happens when cell culture stocks are thawed from -80 degrees. Cells are obviously highly complex nanoscale machines. If they are not frozen properly in the correct freezing media then thawing them will wreck all of their sub-cellular components. This type of phenomenon limits you to:

- Designing machines that work from ultra-cold temperatures all the way up to room temperature (unlikely)
- Designing products that assemble into their final configuration as they thaw (this type of complexity would present a near insurmountable problem)
- Designing incredibly simple products (takes away the point of using a molecular scale assembler)
- Design incredibly complex products capable of fixing themselves (harder than all three of the above)

This type of concept also suffers from the same tool-tip problem of the assembler we discussed above. Atoms don't just stick together like Lego, mechanosynthesis principles dictate that you need specific environmental conditions and specific molecular configurations to break and make bonds. So you would still have to design a system that does this but now at ultra low temperatures.

Now AFM tool tips can be used to move around individual atoms and molecules but the speed at which they do this is abysmal i.e. in real time. Just to illustrate how fast a machine would have to work lets propose that you want to make something very simple: a diamond disk 5cm x 5cm x ~2mm (one chemical mole). If your tool-tip places a trillion atoms a second (bearing in mind you have to do this at near zero, have to have the correct tool-tips and still have to pick up the atoms from a feedstock) it would take nearly 18,000 years. You might be able to speed it up by using multiple tool-tips (like IBMs Millipede AFM project) but it's still not going to be quick.

Lastly aside from those huge technical hurdles you have all the annoying details like how to make an appropriate feedstock, what to do if the machine breaks, how to build these atomic tool tips if you are able to design them, how to create such low temperatures, how to store the data for a template (1 byte per atom for our disk would require 600ZB, 1200x greater than the entire internet) etc etc etc. In summary the scientific and technical requirements for such a project are orders of magnitude greater than what we are capable of today.

If you are interested in real life areas of research that work on similar things I would suggest looking into 3D printers and Fablabs. They have far more promise towards creating portable factories than molecular assemblers do.

hammertime

Well what if we cryogenically froze a human, determined the position of all the atoms in the resulting structure (assuming we overcame all the technical hurdles), then, using that information, used molecular assemblers to build a copy. Would that be possible?

Ryan_m_b

Hammertime did you read my post Yes if we ignore the realities and assume we can do something then hypothetically we can do it but that doesn't really tell us anything.

However if you cryogenically froze a human to the temperature you are talking about you would destroy them, the freezing process would wreck their sub-cellular components and turn their brain to mush.

hammertime

Well, first of all, I should point out that the technical hurdles involved in cryonics are being actively worked on by companies like Alcor. So the realities are not absolute. Cryogenic freezing and reanimation is not impossible like FTL travel or circumventing the Heisenberg Uncertainty Principle. That's why I'm asking you to assume that we were able to perfect cryonics.

What temperature would be needed for us to determine the position of each atom and build a replica with nanobots? Is this temperature lower than the temperature used in cryogenic freezing, and why would it wreck their sub-cellular components?

Frozen is frozen, right?

Ryan_m_b

I'm well aware that cryogenic freezing is being worked on but be wary. The fact that something is being worked on makes it neither possible nor on the way. I'm also highly sceptical of Alcor, their "FAQ for scientists" reads like a high school project rather than a proper summery of the science for fellow experts. For one thing they have no demonstrable knowledge of what nanotechnology actually is, they just assert that with it in situ whole body regeneration will be possible. This is nothing to do with the real life field of research and is everything to do with science fiction.

Humans are mostly water, water freezes to ice, ice is a different density to water. Ice formation wrecks sub-cellular components the same way your skin and organs would be wrecked if I inflated all the water in you. A way around this would be to attempt vitrification but there are still large sub-cellular changes such as deformations, phase transitions etc.

Determining positions of atoms is only possible with something like an electron microscope or atomic force microscope. Both of those would only tell you the position of the atoms on the surface and the temperature would have to be near absolute zero. You would have to section the object bit by bit, it would take thousands - millions of years to "scan" a human atomically and as I pointed out above you need 1200x the storage capacity of the internet to store 12grams of carbon data.

Please, before you start posting about nanobots go and find out what nanotechnology in medicine is really about. I recently blogged about this and included some links. The field of nanotechnology is not about building tiny robots. What you are asking for there is cell sized molecular assemblers which have all the problems I've outlined above plus many, many more.

hammertime

Really? I checked Wikipedia and it said that vitrification has been used on several different types of tissues, including a rabbit kidney that was later successfully transplanted.


Sectioning the object isn't an issue if you want to build replicas. And why would scanning take thousands - millions of years?

hammertime

Also, why couldn't 3-D printers be used to create identical replicas of objects?

Ryan_m_b

And that's a world away from successfully vitrifying a person and resuscitating them. I've yet to see any evidence that brain structure at a molecular level can be preserved, more likely you would cause alterations that at the very least wipe out the resident personality. Lastly: how many kidneys completely failed do you think?
For the reason I outlined above. If you want atomic resolution you need something like an atomic force microscope. These take seconds-hours to scan a 2D area measured in nanometres. You're talking about using bulk products, this will take a lot of time.
It depends on the sophistication of the 3D printer. Most can only make things out of one material for example so a 3D plastic printer is going to be useless at making a metal object. However I suspect that as 3D printer technology get's better we'll start to see a fusion of different approaches so that the printer will have multiple tools for multiple materials e.g. a plastic extruder alongside a metal sintering head.

hammertime

Sure, you haven't seen any evidence now. But the broader point is that we once thought that it would be impossible to vitrify a bodily organ. Then we found out it wasn't. That's proof of concept right there. Sure, it'll take time to find a way to vitrify brains, but there's no reason why it should be impossible. All it takes is one scientist having a "Eureka!" moment.


But is that a fundamental limitation? Is it impossible for them to speed it up to the point where determining the position of every atom in, say, a car would only take a few hours?


Okay, so wouldn't that enable us to create replicas of macroscopic objects at the atomic/molecular level?

Ryan_m_b

It is not proof of concept. Organs are very different, just because there has been success with one does not mean the method is viable elsewhere. In addition the brain is highly reliant on very specific cellular and molecular connections to generate the resident mind. These experiments do not show that a brain could be preserved and resuscitated with the personality intact. Also you suggested this because you want to "scan" a brain and thus want it at near absolute zero, again this is a completely different field to cryopreservation.

You should also be aware that science does not work in the way you describe. The vast majority of the time new technologies come about as the result of meticulous incremental developments. We do not suddenly have one discovery that enables and entire field of technology to pop up fully formed.
The last time I studied the mechanisms of AFM was over a year ago so I'm quite rusty. I do however remember that there are great limits to how fast it can go (there being a minimum time for the reading to take place). Also AFM only reads the surface layer of atoms and there are limitations as to whether or not it can tell what type of atom it is. There is no current technology that allows AFM tooltips to pick up and drop any kind of atom in any configuration. What you are asking for here is not a scaled up version but a totally different albiet similar in principle technology. If you want to learn more about AFM I suggest you post a thread elsewhere or go and learn about it yourself.
No because 3d printers do not work at atomic/molecular resolution and even if they did you would have to have an atomic/molecular scale blueprint of the object in question anyway.

hammertime

I'm not a neuroscientist, but I'm pretty sure that personality and memories are simply the result of a certain configuration of atoms in the brain, not unlike a hard drive. If that's the case, why would vitrifying it and the resuscitating it be any different from doing the same with a kidney?

You're right. But IIRC, we already have some sort of primitive molecular assemblers that are used to create simple microscopic structures. This could be considered analogous to the Wright brothers' plane at Kitty Hawk. It won't be easy to scale up to a utility fog<http://en.wikipedia.org/wiki/Utility_fog>, but we'll eventually get there. Yes, there are significant technical hurdles to overcome, but this is a technological holy grail. I'm confident that we'll one day get to the point where Star Trek replicators are a commonplace item.


Why is there a minimum time for reading to take place?

hammertime

I have a few more counterpoints for this post.

Well, humans didn't evolve from scratch. Natural evolution created simple organisms that gradually got more complex. Technological evolution will do the same for molecular assemblers, albeit at a much faster rate.

Keyword: could. There's certainly no guarantee that it'll be like trying to build a car with ever-changing play-doh.

3) Dealing with wear and tear -
If you have a small error it's possible that the system will break, if your atomic tool-tip moves wrongly (perhaps because of temperature fluctuations) and binds a piece of machinery to a chemical it's dealing with then how do you fix it? You have to design the system to reabsorb and rebuild itself and design it so that this repair system is more efficient than the rate of damage. Considering the complexity of molecular repair faculties in biological systems we are back to square one.

To sum up I don't think there is a legitimate argument that such a thing could not exist however the complexities involved are fantastic. You have to design a system far more complex than all the life that has evolved on Earth.[/QUOTE]

We've already designed several systems far more complex than anything that has evolved naturally: space shuttles, satellites, TV's, computers, video game systems, etc.

Molecular assemblers are still technically possible. Therefore, there will come a day when they are invented. Maybe not in the next decade or anything, but certainly within the next few centuries.

BTW: Sorry if I sound pig-headed. I just have an active imagination and have a hard time believing people when they say "such-and-such will never be possible or is highly unlikely".

Ryan_m_b

Vitrification does not keep atoms locked in the same place. It holds chemical reactions at a very low level of activity, the vitrification and thawing process both cause changes in the cellular and subcellular components. The majority of cells survive this process but they do not come out completely identical to how they went in which would be a must for the brain.
You are imposing an analogy here, I see no reason why it should be considered the same. Fire could be considered a molecular assembler, after all it is used to change the molecular configuration of objects. This may sound facetious but what you are asking for is further beyond current technology as a laptop is beyond a stone age axe.
IIRC a whole bunch of reasons. Essentially everything about an AFM is limited. Why don't you go and read up on it yourself.
So you are proposing using genetic algorithms to invent a molecular assembler? You realise you need to have software and hardware capable of simulating every atom in the machine in real time? Not to mention the environment you are housing it in. Lets run some numbers:

Assuming a density comparable to living organisms ~1kg/L and an assembler made primarily of carbon. Just to simulate a coke can sized assembler would require 132YB of data to just hold the template, i.e. not even run the simulation (generously assuming 1 byte per atom). This is 2000x times greater than the entire internet. In response to this please do not try to cite Moore's law. At the most we get to enjoy about ten more years of that before it reaches the fundamental limitations of nanolithography.

Even if you did have the required data storage, computation and software to simulate thousands of generations of assemblers through genetic algorithms there's no guarantee you will get there. More likely you will just go down an evolutionary dead end and have a useless product.
I'm sorry but no. No chemical just sits there as it is pulled apart, just go and look into protein folding if you want an idea of how composition effects topology.
This really did make me laugh! If you think these trivial machines are more complex than any organism you clearly display a huge ignorance of biology. I don't mean this offensively but honestly, no human invention has ever matched even a simple cell. Go and look into genomics, epigenomics, proteomics, transcriptomics and metabolics.
Two things here: firstly the face that something is technically possible has little bearing on whether or not it will be built. Secondly guessing dates is nonsensical because you cannot have any basis for what you claim.
I have never claimed it is impossible. What has happened throughout this entire thread is that you have asked questions, I have answered them to the best of my ability and knowledge. My answers have dealt with current science and nothing else. I have never indicated that such devices would never exist nor have I indicated a personal opinion on the subject. It is very arrogant for you to dismiss the work of highly intelligent and capable people as well as entire fields of study purely because you personally don't like (or understand) the answers. It is also highly discourteous of you to keep posting questions whilst ignoring the vast majority of answers given to you. You have not demonstrated any will to learn, only to hear the answers you want. This is very frustrating when all I have tried to do is help you learn the real subject rather than at meat to your fantasy.

hammertime

I did. I don't see any fundamental reasons why these limitations can't be overcome. Do you?