Thermally driven self assembly

From apm
(Redirected from Thermally driven assembly)
Jump to: navigation, search

Thermally driven assembly is NOT synonymous to self assembly!!
Nonthermal self assembly is a form of selfassembly that is not thermally driven.

Thermally driven assembly is also called called self assembly or brownian assembly (seldom)
(TODO: Add minimal definition)

  • Today thermally driven assembly is already extensively used (e.g. structural DNA nanotechnology) this will continue onward into the early stages of the development of atomically precise manufacturing (APM).

Beside the actual function of the building block (structural element / machine element) completely unguided thermally driven assembly requires the building blocks to be have a unambiguous unique puzzle piece shape that completely determines its target position. (everything that can stick will stick)

  • Brownian assembly is generally slower then advanced directed assembly like mechanosynthesis. (numbers needed)
  • The ambient temperature dictates diffusion speed.
  • Lower dimensionality that is diffusion on a surface instead of a volume or on a line instead of a surface speeds up the process.
  • Dividing one long diffusion path to several shorter irreversible diffusion transport stretches speeds up the process.

Possible sub-classifications

Self connectedness

(wiki-TODO: what are the overlaps and differences)
(wiki-TODO: what are advantages and limitations)

Building block type

  • small molecule monomers (normal chemistry)
  • floppy chains (foldamer chains)
  • rigid blocks (sufficiently rigid to hold their shape)

Selfassembly levels

chemical monomer addition floppy chain rigid block
folding impossible
small monomers hardly
can fold up on themselves
the folding up of foldamers
like e.g. in protein folding

Ribosomes have this case on their output side.

the folding up of hinged chains of stiff blocks
utilization of longer range forces allows
for nonthermal self assembly

Ribosome like chain assembly of bigger blocks
would have this on the output side

one pot
not too useful
simple monomers do not carry enough
position encoding structure
to self assemble anything but
unconstrainedly growing crystals

Adding some positional capabilities one gets to:
patterned layer epitaxy which is a form of
site-specific workpiece activation

Ribosomes have this on their input side (sort of).
Amino acid monomers are delivered on handles to encode
their identity in an a more easily decodable form
(the digital DNA code)

structural DNA nanotechnology
with short DNA oglionucleotides as "bricks"
thermally driven one pot self assembly
on the second self assembly level

Adding some positional capabilities one gets to:
patterned block epitaxy wich is a form of
Site-specific workpiece activation
and to: Tether assisted assembly

Ribosome like chain assembly of bigger blocks
would have this on the input side

synthesis of foldamers from monomers
(often done with beads as starting seeds)
(??) for iterative addition of longer chains
interactions must not be too strong
to avoid kinetic traps
thermally driven iterative self assembly
on the second self assembly level

(wiki-TODO: illustrative icons would greatly help in making this table much more comprehensible)


What is progress in self assembly capabilities

  • Advancednedd of self-assembly is about address space size
  • Advancedness of self assembly is not about plain size

Size of self assembled structure is not a good measure for the advancedness of ones artificial self assembly technology. The same goes for self assemblies that go around 360° full circle.

When all parts have an identical self matching shape then this gives ride to infinite translatory or rotatory symmetries.

What really counts for judging the advancedness is the size of fully and arbitrarily addressable space.

In structural DNA nanotechnology

First level self assembly of DNA staple bricks to DNA blocks:

  • Yields drop quickly with size
  • addressable space is finite due to staple strands of suitable length only capably of carrying a few bits of info

Second level self assembly of pre-self-assembled DNA blocks to multiblock structures:

  • Bigger blocks diffuse more slowly

In case of 3D DNA blocks there is redundant access self-assembly involved contributing to the robustness of the process.

In case of de-novo protein engineering

Of most interest as basic reusable building bricks are
the most predictably folding and most robust protein structures
that come closest to prismatic brick shapes. Helices and sheets.

What one wants is both:

  • high specificity (wrong interface pairings don not stick and hold together well)
  • high affinity (right interfaces do fit and hold together well)

What one wants to design is a multi-orthogonal set of interface pairings.
One can chart a matrix listing the strengths of all possible interface pairings.

In case of proteins the side-chains already fulfill the purpose of defining the proteins shape (how the backbone is supposed to fold) Redirecting too many of them to define an outward facing inter-protein interface instead can start destabilizing the proteins shape itself.

Potentially destabilizing factors:

  • putting too many side chains to use to define an outerinterface
  • putting too many side chains of similar polarity adjacently
  • putting too many hydrophobic side chains outward
  • putting too many side chains to other use (for some specific applications)

All together this results in that (in the case of proteins) the size of the sets of achievable "good" orthogonal interfaces is rather very small. Countable on one hand.

So only very small assemblies can be achieved via one-pot assembly. For a bit bigger addressable spaces iterative self-assembly is needed. This cannot be done directly in protein synthesizing cells though, which would provide some protein folding help (via the chaperone folding helper proteins).

In iterative self-finding self-assembly of protein blocks the assembly happens in a linear that is branchless sequence. In the case of protein blocks the product structure is not necessarily geometrically linear and not necessarily flexible).

In a linear assembly sequence failure rates multiply with each iterative self-assembly step. Much like failure rates multiply up in the case of the synthesis of foldamer chain molecules. So even iterative self assembly hits its limits soon.

Redundant access self-assembly seems to be able to avoid the problem with the multiplying error rates. But this would probably need some amount of one pot assembly? (TODO: investigate here)

Maybe maybe enough structure could be encoded via 2nd level iterative self assembly such that a 3rd level one pot self assembly becomes possible.

Foldamer in foldamer inset:

Given all this trouble with the scaling de-novo protein self-assembly outlined above: It might be easier to integrate/embed just somewhat scaled de-novo protein engineering into a stuctural DNA nanotechnology framework surrounding it. Compensating the massive drop in stiffness by a sufficiently large interface area between the foldamer technologies.


External links

There's huge amount of literature on thermally driven assembly.