[ExI] [Hplusroadmap] Fwd: Alternate transistors

Bryan Bishop kanzure at gmail.com
Tue Feb 26 04:17:14 UTC 2008

This might be of interest re: self-replication. Given easy-to-synthesize 
transistors, RepRap can start doing self-repping.

- Bryan

----------  Forwarded Message  ----------

Subject: Alternate transistors
Date: Monday 25 February 2008
From: Bryan Bishop <kanzure at gmail.com>
To: Blake Rieger <biomors at gmail.com>

The LiquiFET was using capillary action to work. Wouldn't scale up to my 
transistor-in-a-jar idea for testing. Only nm. Maybe we can come up 
with another way to make it happen.

Really, for digital logic, all we need is a relay (switch). But the 
problem is that if you have 20 relays connected in sequence, you get 
voltage drop-off eventually (obviously) and the circuit just fails 
completely. The trick with a transistor is that it acts as an amplifier 
too, keeping the voltage always at the binary logic state. State 

In field effect transistors (jFETs), the voltage to the gate pin 
determines the magnitude of current flowing through the circuit via 
directly influencing the 'resistance' of the channel. As the gate 
voltage increases, the depletion region increases in size until 
eventually there's zero current allowed to flow from source to drain. 
The depletion region will not carry current. This works because of the 
purities of the silicon material and the doping. And it works because 
the electric field naturally repels electrons, constricting the current 
(= generating resistence). I think it's pretty much an example of two 
differently sourced junctions (so that one stream of electrons want a 
ground, the others (holes?) want to go to a source) or something like 
that. If we wanted to do this without semiconductor manufacturing, we'd 
need a makeshift semiconductor-- where are we going to find one of 
these? Actually, we just need a digital relay. But first.

There's one way to do it: piezoelectricity. Have a metal piezo contact 
that would complete the circuit. Applying voltage to the piezo means we 
can change the metal's shape. This would be useful for, say, 
compressing a liquid, which when compressed would then transmit its own 
signal (input/output of its own). This *might* scale well for working 
with our hands -- get a piezo from a speaker, maybe, but what liquid 
would experience such a drastic change in conductivity at room 
temperatures? Solid state is a much, much better idea.

Most mechanical relays are clunky and slow. A piezo means we can change 
the shape of the metal and make it an open/closed circuit within 
picoseconds of applying the voltage, so maybe the mechanical idea is 
not so bad after all? I tried running some searches on the net for 
piezo transistors and piezo relays. Nothing out there yet.

There's a company (Kovio) that uses nanocrystals (pre-doped si) to do 
inkjet printing of transistors. They claim to be going to market later 
this year. I am wondering how they are generating their pre-doped 

Doped semiconductor powder and the preparation thereof:

There happens to be a few PDFs that Google can find on the net for the 
synthesis of silicon nanocrystals. Can we figure out the synthesis of 
doped si nanocrystals?  One of the 2001 papers I've looked at was using 
UHV for si nanocryst synth -- seriously not good. UHV bad. Not simple. 

There's a paper out there called "Synthesis of silicon nanocrystals in 
aluminum-doped SiO2 film by laser ablation method." That might be 
interesting. Can you find it for me?

Ah, here's a method that requires just a solution to do the synth:

This looks interesting too:
> Kortshagen and co-workers begin by injecting a dilute mixture of 5%
> silane (SiH4) in 95% helium and argon into a narrow quartz tube about 
> 23 centimetres long. They then apply around 200 watts of power at a 
> frequency of 13.56 megahertz to a ring-like electrode that is about 10 
> centimetres away from the ground electrode. The resulting plasma is 
> unstable and is made up of a filament of bright plasma globules. 
> Existing approaches to plasma synthesis use stable, uniform plasmas.       
> The high-energy electrons in the plasma decompose the silane gas into
> its constituents and the silicon atoms released in this way then 
> recombine to form silicon particles. Transmission electron microscopy 
> reveals that the nanoparticles are all between 20 and 80 nanometres in 
> size and are predominantly cubic-shaped.    

Most of the solution-based si nanocryst synth methods seem to involve 
something known as a Zintl precursor, and it's actually done in silicon 
tetrachloride solution or something, so it doesn't sound like anything 
too hideously complex has to be used for the wet-based synthesis of 
nanocrystals. If we can come up with a si nanocryst powder, we can use 
this in an inkjet printer, print out the relevant circuits, and there 
we go. Print out relays or transistors, probably. How hard is it to 
work with silicon tetrachloride, anyway? 

Synthesis of semiconductor nanoparticles.pdf
>         The synthesis of semiconductor nanocrystals has been well
>         developed for II-VI semiconductors (CdSe, CdS, ZnSe, ZnS, 
>         CdTe) and III-V semiconductors (InP, InAs, GaAs) [1, 2]. The 
>         development of the pyrolytic route to making nanocrystals was 
>         one of the major advancements in the nanocrystal synthesis, 
>         and was the first type to produce samples with size 
>         distributions that were reasonable. Large quantities (≤ 1 
>         gram) of nanocrystals can be produced in a few hours. The 
>         underlying strategy of this synthesis is the reaction of the 
>         molecular precursors instantaneously at high temperature in a 
>         coordinating solvent. Varying the concentrations of the 
>         reagents with respect to the reaction solution and growth time 
>         controls of the size of the particles.            

* solution phase synthesis route

>        Synthesis details         II-VI (CdSe)
> Dimethyl cadmium is obtained commercially and transferred into a vial
> by a vacuum transfer under Ar at a pressure ≤ 20 mtorr. This removes 
(how do we do vacuum transfers??)
> impurities which have a higher vapor pressure than the pure compound. 
> Pure Se metal is purchased under Ar and dissolved in tributyl 
> phosphine in the glove box. The dimethyl cadmium is added such that 
> the molar ratio between Cd:Se is 1:1. Size and shape control are the 
> result of variations in the concentration. The solution should remain 
> clear, indicating that the reagents have not reacted. 
> Trioctylphosphine oxide (90%) (TOPO) is degassed and heated under Ar 
> to 360°C. The impurity in TOPO is crucial to the stabilization of the 
> nanocrystals in solution [3]. The stock solution is injected 
> immediately while the temperature is simultaneously lowered to the 
> temperature for nanocrystal growth (300°C). Further injections will 
> boost the size as well as improve the size distribution [2]. During 
> the synthesis, the absorption spectrum is monitored to ensure that 
> particle growth does not stop due to consumption of the reagents in 
> solution. Once particle growth stops, Ostwald ripening begins, which 
> is the growth of larger particles at the expense of smaller ones, an 
> undesirable process as it increases the size distribution. After the 
> reaction is complete (either when the desired size is reached or the 
> reagents are all consumed), the temperature is lowered and the 
> nanocrystals are precipitated at room temperature with MeOH under Ar 
> or air. The particles as synthesized are soluble in toluene, hexanes, 
> and pyridine and can be stored as a solution or a powder indefinitely 
> in dark and under Ar. The TOPO on the surface can be exchanged with 
> another ligand or removed to change its solubility properties by 
> refluxing with a labile ligand such as pyridine.                          

So this looks like a good method to pursue. Kovio already makes circuits 
for themselves w/ transistors. So I'm betting that they have hired on 
one of the grad students that worked on nanocrystal synth since 1993 or 
whenever the first observation was made of their general production. I 
doubt they have their own too heavily customized protocol for the 
generation of nanocrysts. So if we start collecting documents on the 
synthesis of nanoparticles, start to look for commonalities, and also 
trying to find ways to reduce the needed equipment to stuff that the 
RepRap should be able to make on its own, we should be set for full 
self-replication. Although I was really hoping for the transistor in a 
jar to work out for the better ... maybe just a relay in a jar? Please?

- Bryan

Various stuff:

* Growth of Highly Luminescent Silicon Nanocrystals by Rapid Thermal 
Chemical Vapor Deposition

Bryan Bishop

More information about the extropy-chat mailing list